THE PSYCHOLOGY OF CHESS

Do you need to be a genius to be good at chess? What does it take

to become a grandmaster? Can computer programmes beat human
intuition in gameplay?
The Psychology of Chess is an insightful overview of the roles of
intelligence, expertise and human intuition in playing this complex
and ancient game. The book explores the idea of “practice makes
perfect”, alongside accounts of why men perform better than women
in international rankings, and why chess has become synonymous
with extreme intelligence as well as madness.
When artificial intelligence researchers are increasingly studying
chess to develop machine learning, The Psychology of Chess shows us how
much chess has already taught us about the human mind.

Fernand Gobet is Professor of Decision Making and Expertise at the

University of Liverpool, UK. He is a chess international master, and
played numerous times for the Swiss national team.
THE PSYCHOLOGY OF EVERYTHING

The Psychology of Everything is a series of books which debunk the myths

and pseudo-science surrounding some of life’s biggest questions.
The series explores the hidden psychological factors that drive us,
from our sub-conscious desires and aversions, to the innate social
instincts handed to us across the generations. Accessible, informative,
and always intriguing, each book is written by an expert in the field,
examining how research-based knowledge compares with popular
wisdom, and illustrating the potential of psychology to enrich our
understanding of humanity and modern life.
Applying a psychological lens to an array of topics and contemporary
concerns – from sex to addiction to conspiracy theories – The Psychology
of Everything will make you look at everything in a new way.

Titles in the series

The Psychology of Grief
Richard Gross
The Psychology of Sex
Meg-John Barker
The Psychology of Dieting
Jane Ogden
The Psychology of Performance
Stewart T. Cotterill
The Psychology of Trust
Ken J. Rotenberg
The Psychology of Working
Life
Toon Taris
The Psychology of
Conspiracy Theories
Jan-Willem van Prooijen
The Psychology of Addiction
Jenny Svanberg
The Psychology of Fashion
Carolyn Mair
The Psychology of Gardening
Harriet Gross
The Psychology of Gender
Gary W. Wood
The Psychology of Climate Change
Geoffrey Beattie
The Psychology of Vampires The Psychology of Driving
David Cohen Graham J. Hole

The Psychology of Chess The Psychology of Retirement

Fernand Gobet Doreen Rosenthal and Susan M. Moore

The Psychology of Music The Psychology of School

Susan Hallam Bullying
Peter Smith
The Psychology of
Weather The Psychology of Celebrity
Trevor Harley Gayle Stever

For further information about this series please visit:

www.thepsychologyofeverything.co.uk
THE PSYCHOLOGY
OF CHESS

FERNAND GOBET
First published 2019
by Routledge
2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN
and by Routledge
711 Third Avenue, New York, NY 10017
Routledge is an imprint of the Taylor & Francis Group, an informa business
© 2019 Fernand Gobet
The right of Fernand Gobet to be identified as author of this work has
been asserted by him in accordance with sections 77 and 78 of the
Copyright, Designs and Patents Act 1988.
All rights reserved. No part of this book may be reprinted or
reproduced or utilised in any form or by any electronic, mechanical,
or other means, now known or hereafter invented, including
photocopying and recording, or in any information storage or
retrieval system, without permission in writing from the publishers.
Trademark notice: Product or corporate names may be trademarks or
registered trademarks, and are used only for identification and
explanation without intent to infringe.
British Library Cataloguing-in-Publication Data
A catalogue record for this book is available from the British Library
Library of Congress Cataloging-in-Publication Data
Names: Gobet, Fernand, author.
Title: The psychology of chess / Fernand Gobet.
Description: New York : Routledge, 2019. | Series: The Psychology
of Everything
Identifiers: LCCN 2018020887 (print) | LCCN 2018024430 (ebook) |
ISBN 9781315441863 (ePub) | ISBN 9781315441870 (Adobe) |
ISBN 9781315441856 (Mobipocket) | ISBN 9781138216631
(Hardback) | ISBN 9781138216655 (Paperback) |
ISBN 9781315441887 (eBook)
Subjects: LCSH: Chess—Psychological aspects.
Classification: LCC GV1448 (ebook) | LCC GV1448 .G634 2019 (print) |
DDC 794.1019—dc23
LC record available at https://lccn.loc.gov/2018020887
ISBN: 978-1-138-21663-1 (hbk)
ISBN: 978-1-138-21665-5 (pbk)
ISBN: 978-1-315-44188-7 (ebk)

Typeset in Joanna
by Apex CoVantage, LLC
 To the Queens of my Life

Chess is like life.

World champion Boris Spassky

Chess is life.
World champion Bobby Fischer

Chess is not life.

World champion Hou Yifan
 CONTENTS

Preface xi
Acknowledgements xiii

Opening 1

1 The eye of the master 7

2 Chunks! 13

3 The best move 29

4 Practice makes (almost) perfect 43

5 Men vs. women 55

6 Style and intuition 65

7 Errare humanum est 77

8 Psychological warfare and training techniques 89

x CONTENTS

9 The magic bullet? 101

10 Costs of playing chess 107

11 Endgame 117

Further reading 119

 PREFACE

This book is a non-technical introduction to the psychology of chess.

It is a book I thought of writing many years ago. Thirty years ago,
to be precise, when I started working on my PhD on chess players’
memory. At the time, I decided that not enough was known about the
topic to write such a book, and that I should wait a few more years.
The wait took longer than expected. Although I did write three
monographs on chess memory and board games, and two intro-
ductions to the psychology of expertise, I never went back to the
original idea.
When Ceri McLardy invited me to write a short book on the topic
for the series The Psychology of Everything, I accepted immediately – I
could not wait any longer! Besides, now we do know quite a lot about
the psychology of chess.
The book is aimed at three main audiences. First, undergraduate
students. For this audience, general implications for psychology will
be drawn from chess research. Second, general readers interested in
the psychology of chess. Several of the selected themes will be of
directed interest to them. Third, chess players who want to under-
stand the psychological mechanisms underpinning the way they
think when playing chess. Sections on errors, style, intuition and
training will be of particular interest to them. Knowledge of the rules
xii P R E F A C E

of the game is not necessary to enjoy this book, and care has been
taken not to use chess jargon, or to explain it when it is used. No
background in psychology is expected, either.
In line with this collection, the number of references is kept to
a minimum. The suggested items in the Further Reading section at
the end of the book provide sources where detailed pointers to the
literature can be found.
Around the time of my PhD, I had the very good fortune of col-
laborating with Adriaan de Groot and Herbert Simon, two outstand-
ing scientists who made seminal contributions to the psychology
of chess. Later, I trained several gifted researchers (Merim Bilalić,
Guillermo Campitelli, Philippe Chassy and Giovanni Sala) who had
decided to devote their PhD research to aspects of chess psychology.
Both the old and new research will be described in this book.
In my first career, I was a professional chess player, reaching the
title of an international master and playing for the Swiss national
team. Therefore, I have sometimes taken the liberty of providing first-
hand evidence and insights about chess players and their mind.
 ACKNOWLEDGEMENTS

Many thanks to Guillermo Campitelli, Philippe Chassy, Morgan Ereku,

Alexey Root and Giovanni Sala for their comments on a draft of this
book and many exciting discussions about chess psychology.
 OPENING

Chess fascinates many people, even those who do not play it. It is
played at all ages, in all countries, without any attention to religion
or ethnic background. As a symbol of intelligence and good decision-
making, it is often used in advertisements for banking and business.
As a symbol of genius bordering on madness, it is a common theme
in fiction and has generated numerous books and movies.
Born in India in the 6th century, chess was adopted in Persia (Iran)
in the next century, and then by the Arabs. The first books on chess
were written in the 9th century, but did not survive to our day. It is
also in the 9th century that chess was introduced to Europe, through
Spain and Sicily. By the 13th century, chess was the dominant game
in Europe, as witnessed in the Book of Games, written in 1283 at the
request of Alfonso X of Castile: “Since chess is the noblest game,
which requires the most skill compared to all the other games, we
are going to talk about it first of all”. By then, its influence had spread
well beyond the sphere of games and it was part of culture at large.
For example, in the second half of the 13th century, Jacobus de Ces-
solis, an Italian monk, preached morality using chess as metaphor:
“In life, as on the chessboard, each piece has its own rights but also
its own obligations”.
2 OPENING

Chess has sometimes been the mirror of historical developments.

At the end of the 15th century, at a time when women were gain-
ing importance in politics in medieval Europe, the rules of the game
changed, and the until-then modest vizier – who could move only
one square diagonally – mutated into the powerful queen, who can
move in any direction without any limitation. Anticipating the French
Revolution by 40 years, François-André Danican Philidor, the best
player of the time and an opera composer, asserted that the pawns
(the weakest of the chess pieces) were not just cannon fodder, but
actually were the “soul of chess”. This paved the way for the scientific
approach to chess, and in particular the theory developed by Wilhelm
Steinitz. The impact of his approach on chess has been compared to
the impact of Newtonian theory in physics.
At the height of the Cold War, the match between American Bobby
Fischer and Soviet Boris Spassky captured the imagination of the pub-
lic. So did the matches between political refugee Viktor Korchnoi
and Soviet apparatchik Anatoly Karpov, spiced up by various scandals
including the presence of a parapsychologist on Karpov’s team alleg-
edly hypnotising his opponent.
In 1997, world champion Garry Kasparov defended the honour
of the human race against artificial intelligence in his match against
Deep Blue – and lost. This defeat, a milestone in the history of science
and technology, led to considerable soul searching in the media and
was the inspiration behind several books and movies.

CHESS AND SCIENCE

The nature of chess has been often debated in the literature. It is fre-
quently presented not only as a game, but also as a sport (because of its
competitive element), an art (because of the beautiful combinations
it allows) and a science (because of the systematic way it is studied).
Although chess players speak of chess “theory”, the term is not
used in the same way as in science. Rather than a formal system of
laws, mechanisms and principles, “theory” in chess means a catalogue
of initial moves and their evaluation for the theory of openings, and a
 OPENING 3

discussion of key positions and the way to handle them for the theory
of endgames. There were some attempts to identify the fundamental
principles of play, most notably by Wilhelm Steinitz, the founder of
the classical school of chess, with his 1889 textbook The Modern Chess
Instructor, and Aron Nimzowitsch, one of the founders of the hyper-
modern school, with his 1925 book My System. However, despite their
great originality, both fell short of the rigour of a scientific theory. In
addition, what is lacking generally with chess is any effort to test its
“theories” in a systematic way, beyond recording new games. Such
tests are obviously at the heart of scientific research.
Chess has been the topic of much scientific research. It has been
investigated by a number of academic disciplines, including sociol-
ogy, ethnology, philosophy, mathematics and neuroscience. By far, it is
in computer science (including artificial intelligence) and psychology
that chess has been studied to the greatest effect. In artificial intelli-
gence, chess has been a standard task for the development of machine
learning and search algorithms. In psychology, it has been the topic
of seminal research into perception, memory, learning, thinking and
decision-making. It has sometimes been called the drosophila of cogni-
tive psychology, by analogy to the role of the fruit fly in genetics.

A RICH DOMAIN OF RESEARCH FOR

PSYCHOLOGY
Starting with French psychologist Alfred Binet (the creator of the
first successful test of intelligence), chess has excited the imagination
of psychologists. Binet studied blindfold chess, a variant of chess
where players do not see the board. In 1925, a group of Russian
psychologists took advantage of the Moscow tournament, which
brought together the best players of the time, to administer a battery
of psychometric tests, measuring abilities such as memory, intelli-
gence and even motivation with the Rorschach test. They found that
there was hardly any difference between chess masters and a control
group matched for intelligence, except for tests using chess material
and tests measuring the ability to distribute attention and discover
4 OPENING

logical principles. They also provided a list of the physical and mental
qualities required by chess, which were instrumental in convincing
the Soviet government that chess should be encouraged as an activity
leading to the development of self-discipline and the improvement
of intellectual competences.
However, it is to Dutch psychologist and chess master Adriaan de
Groot that we owe the first experimental study on chess psychol-
ogy, which he carried out for his PhD research. In 1938, he was
earning money as a journalist by covering the AVRO tournament in
Amsterdam, which brought together the world’s best eight players. He
managed to convince five of the participants to take part in his experi-
ments, including world champions Alexander Alekhine and Max Euwe.
Amusingly, some of the data were collected after the tournament on the
steamer carrying many European masters to Buenos Aires, where the
1939 Olympiads (world championship by teams) were held. As the trip
was rather long, players were grateful to participate in these experi-
ments and therefore to break the monotony of the voyage.
De Groot studied not only chess players’ ability to find good
moves, but also their ability to rapidly understand the gist of a posi-
tion even after seeing it just for a few seconds, as well as their ability
to memorise these positions rapidly and accurately. Many of the ideas
presented in this book can be traced backed to de Groot’s PhD thesis.
The second key study in chess psychology, carried out by Herbert
Simon and William Chase in 1973 at Carnegie Mellon University in
Pittsburgh, developed a powerful theory – called chunking theory –
to explain de Groot’s data. Combining experimental methods with
ideas from artificial intelligence and computational modelling, Simon
and Chase performed a series of experiments that inspired much of
the research carried out in the following decades.
Nowadays, chess psychology is an active domain of research and
is arguably still the main domain in expertise research. Many dif-
ferent aspects of chess are studied, from cognition to personality to
intelligence. Several reasons explain this popularity, including: chess
has its own rating system, the Elo rating, which offers a precise and
up-to-date measure of skill; it has an ideal balance between simplicity
 OPENING 5

and complexity; it allows many experimental manipulations; and it

has strong external validity. In fact, the key discoveries made in chess
psychology generalise to most domains of expertise, and indeed to
psychology in general, as we shall see. Thus, researchers often study
chess not for its own sake, but for understanding expertise in general.

PREVIEW OF BOOK
The following three chapters cover standard topics in cognitive psy-
chology: perception, learning, memory and decision-making. They
are anchored in the work of de Groot, Simon and Chase. Chapter 4
discusses the relative roles of practice and talent, and emphasises the
importance of studying their interaction. Chapter 5 tries to under-
stand why men appear to perform better than women in chess at the
top level. The following three chapters are devoted to applied chess
psychology, and cover topics such as errors, style, intuition, train-
ing, psychological warfare and cheating. The discussion of style and
intuition will draw on recent developments in artificial intelligence,
focusing on what they tell us about human psychology. Chapter 9
examines the possible benefits of playing chess (e.g. for education or
psychotherapy) and Chapter 10 addresses the potential costs (e.g. the
hypothesis that there is a link between madness and chess).

APPENDIX: A VERY SHORT INTRODUCTION

TO CHESS
In this book, chess refers to international chess, which is distinct
from Japanese chess (shogi) and Chinese chess (shiang qi). Chess
is a game played on an 8 × 8 board. The aim of the game is to cap-
ture (checkmate) the opponent’s king. At the beginning of the game,
each player, called White and Black, has eight pieces and eight pawns.
The White pieces are placed on the first row in the following order:
rook, knight, bishop, queen, king, bishop, knight and rook. The White
pawns are placed on the second row. The Black pieces and pawns are
placed in the same way on the eighth and seventh row. Rooks move
6 OPENING

horizontally and vertically, and bishops move diagonally. The queen,

the most powerful piece, moves horizontally, vertically and diagonally.
The king moves one square in any direction. The knight moves two
squares horizontally (or vertically), and then one square vertically (or
horizontally), thus forming an L shape. Unlike the other pieces, it can
jump over other pieces. If a move ends on a square occupied by an
opponent’s piece, this piece is removed from the board (captured).
Pawns move one square forward (from their starting square, they can
also move two squares forward) and capture one square diagonally.
Additional rules include castling (a move where both the king and
a rook move simultaneously), taking en passant (a pawn moving two
squares from its starting position can be captured by an opponent’s
pawn as if it had moved only one square), stalemate (if one side can-
not move but is not in check, then the game is a draw) and pawn
promotion (when a pawn reaches the end of the board, it may be
replaced by a stronger piece of the same colour).
White moves first, which provides a small advantage. A typical
game can be divided into three phases, although the boundaries
are often fuzzy: the opening (the first moves of the game where
each side develops their pieces), the middle game (where each side
manoeuvres to gain a material or positional advantage whilst thwart-
ing the opponent’s objectives) and the endgame (where few pieces
and pawns remain on the board).
To record games, chess players use the algebraic notation, where
the letters refer to files and numbers to rows. Pieces (except pawns)
are indicated by their initial letter, with N being used for “knight”
to avoid confusion with “king”. For example, the first moves of the
shortest possible game are written as 1.f3 e5 2.g4 Qh4 checkmate.
Chess skill is measured with the Elo rating, a method invented by
Arpad Elo. It is an interval scale with a theoretical mean of 1500 and
a theoretical standard deviation of 200. The following skill levels are
often used (with the Elo range in parentheses): class D players (1200–
1400), class C players (1400–1600), class B players (1600–1800),
class A players (1800–2000), experts or candidate masters (2000–
2200), masters (2200–2400), international masters (2400–2500)
and grandmasters (above 2500).
 1
THE EYE OF THE MASTER

Anybody who has seen chess masters playing bullet chess (1 minute per
side for the entire game) or simultaneous exhibitions, where they play
against 30 or 40 opponents at the same time, would have been struck by
their amazing ability to play good moves very quickly. Indeed, the qual-
ity of moves played under these taxing conditions is surprisingly high,
although not as high as with games played under normal conditions (on
average, 3 minutes per move). It is as if masters see the board differently
than weaker players. Where novices see wooden or plastic pieces, masters
see trajectories, ideas, concepts and sequences of moves. In fact, the same
applies in other fields: one of the hallmarks of experts in science, medicine
and sport is the ability to rapidly perceive the key features of a problem.
The first person to have addressed this question empirically was
Adriaan de Groot in his doctoral dissertation, originally published
in Dutch in 1946 and translated in English in 1965. Because of the
number of issues it addressed and its strong scientific impact, this
work has become a classic in psychology.

A BETTER UNDERSTANDING AFTER 5

SECONDS THAN AFTER 15 MINUTES!
De Groot’s main interest concerned the processes that allow chess
players to choose a move. Specifically, he wanted to test the hypoth-
esis that, compared to amateurs, chess masters considered more
8 THE EYE OF THE MASTER

positions when they were looking ahead and that they anticipated
longer sequences of moves – that is, they were searching deeper. In a
first experiment, he gave chess players a board position unknown to
them and asked them to select what they thought was the best move.
He also asked them to say aloud what they were thinking about.
The players consisted of amateurs, candidate masters and world-class
grandmasters, including world champions. The transcripts of play-
ers’ utterances – called verbal protocols – were then analysed in great
detail, both qualitatively and quantitatively.
The results did not support his expectations: although grandmas-
ters played better moves, they did not differ substantially from other
players with respect to structural variables such as the depth of search,
the number of moves considered or the strategies used when carrying
out search (see Chapter 3 for details). There was an important dif-
ference, however. The best players were able to pinpoint promising
solutions very rapidly, which allowed them to narrow down their
search drastically. As de Groot put it, the world champion understood
the problem position better after 5 seconds than a candidate master after
15 minutes! This was fully unexpected. What was critical was not the
detail of the way players analysed the position by trying out different
moves, sometimes for more than 30 minutes. Rather, the difference
resided in the very first few seconds of seeing a position: perception
is central in chess expertise.
In a second experiment, de Groot directly tested this hypoth-
esis. He presented a position briefly, from 2 to 15 seconds, took it
away from participants’ view, and asked them to reconstruct it as
precisely as they could. As expected, grandmasters did much bet-
ter than candidate masters, who in turn did better than amateurs.
Whilst a grandmaster could reconstruct nearly the entire posi-
tion correctly, a strong amateur struggled to remember half of the
pieces.
De Groot explored several variants of this experiment. In some
versions, he asked players to think aloud, either during the presenta-
tion of the position, immediately after or 30 seconds after. From the
protocols, it is clear that experts did not see individual pieces, but
 THE EYE OF THE MASTER 9

rather saw large complexes, in which perceptual aspects are inter-

twined with dynamic possibilities. In fact, they rarely perceived static
groups of pieces, but almost always threats, probable moves and even
sequences of moves. Note that this is the case even when players are
told explicitly beforehand that the task is to recall a position, and not
to find the best move.
I will have more to say about this experiment in Chapter 2, but
for the time being it is important to realise that this task was in de
Groot’s mind a perceptual task, the aim of which was to understand
what grandmasters saw during the first seconds they looked at an
unknown position. Later on, this task became highly popular and
in more recent research has been predominantly used for studying
memory.

RECORDING EYE MOVEMENTS

A natural way to study skill differences in perception is to record eye
movements. This technology was not available before the war when
de Groot collected the data of his PhD research, and it is only in the
1960s that he was able to carry out such an experiment, with his PhD
student Riekent Jongman. The task was again to reproduce a chess
position presented briefly, this time uniformly for 5 seconds. The data
were fully analysed even later, in a book de Groot wrote with Jong-
man and myself (see Further Reading). There were clear differences
between weak players and masters, with the latter having shorter and
less variable fixations. Masters’ fixations also covered more squares
and landed more often on the squares that were important from a
chess point of view. Another interesting result was that masters fixated
more often on the intersection of squares than the weak players. This
buttresses the hypothesis that masters perceive groups of pieces rather
than individual pieces. Finally, it is likely that fairly simple visual
cues – such as a White pawn that has penetrated Black’s defence –
direct masters’ eye movements to significant squares. In chess, per-
ceptually salient features correlate very often with the strategic and
tactical meanings of a position.
10 T H E E Y E O F T H E M A S T E R

Just like some of the old experiments, the eye-movement experi-

ments used verbal protocols. After reconstructing a position – more
or less successfully – players were requested to retrospect on what
they had seen during its presentation. In general, masters’ retrospec-
tive descriptions broadly agreed with the actual sequence of eye fixa-
tions. An interesting exception was that, in the cases where they had
fixated the same square several times, players tended to remember
only the first fixation. The same finding has been documented in
experiments measuring memory for sequences of words, where
repeated items tend to be recalled poorly – a phenomenon known as
the Ranschburg effect. The protocols were also useful for providing
information about where players directed their attention and about
the way they dealt with atypical positions (e.g. positions that cannot
be put in standard categories).
Later experiments measuring eye movements have also produced
striking skill differences. For example, Charness and colleagues used a
problem-solving task in which participants (intermediate players and
candidate masters) had to find the winning move in a position. In addi-
tion to being faster and choosing better moves, the candidate masters
had fewer fixations but larger eye movements than the intermediate
players.Their first fixations tended to land on empty squares more often,
and, when considering fixations on pieces only, they fixated on impor-
tant pieces more frequently. In general, the data supported the idea that
strong players combined perceptual knowledge with the information
provided by peripheral vision to direct their eye movements.

PERCEPTION: INCREMENTAL AND

ANTICIPATORY
A classic debate in psychology concerns the nature of perception:
is it holistic, with objects perceived in their totality, as maintained
by Gestalt psychologists, or is it constructed by incremental mecha-
nisms, as argued by reductionists? Although not a Gestalt psychologist
himself, de Groot proposed that strong players start with a “land-
scape view” of the board, which provides a global impression of
 THE EYE OF THE MASTER 11

the position, with the details omitted. In the 1960s, there was an
interesting debate about this issue between Soviet psychologist Oleg
Tikhomirov and Herbert Simon, with Tikhomirov defending the view
that perception is holistic, while Simon argued that local mecha-
nisms (e.g. perception of relations of defence and attack between
pieces) were sufficient for explaining the data. More recently, Gobet
and Chassy run computer simulations based on the idea of chunks
and templates (see the next chapter), showing that experts’ percep-
tion, even though it might look holistic, can be accounted for by the
incremental construction of an internal representation using patterns
that are initially fairly small.
Another idea proposed by de Groot seems better supported by the
empirical evidence. He suggested that chess masters used anticipatory
schemas. These dynamic schemas contain information allowing play-
ers to anticipate potential actions. As experts have more and better
developed schemas, they can anticipate actions better. Vincent Fer-
rari and colleagues at the University of Provence (France) tested this
hypothesis. In a first experiment, players saw two positions in quick
succession and had to say whether the second position was the same
as the first one. The results showed that strong amateurs performed
better when the two positions appeared as a normal sequence of
moves, unlike beginners who could not use information about the
normality of moves.
In a second experiment, Ferrari and colleagues studied whether
players tend to recall positions as they were shown or, rather, the posi-
tions that would occur after the standard move is played, as predicted
by the presence of anticipatory schemas. A recognition task was used.
Players saw 10 chess positions displayed in succession; half of the
positions were standard opening situations, while the other half were
a different set of opening situations, this time with one additional
move played. During the recognition phase, players were presented
with 10 old positions (the positions they had seen in the first phase
of the experiment) and 10 new positions (half were the positions
they had seen plus one standard move, and the other half were the
positions they had seen minus the standard move). The results showed
12 T H E E Y E O F T H E M A S T E R

that the strongest players (class A players) made many false recogni-
tions, where they recalled not the position they had seen, but the
position after the normal move had been made. The beauty of this
experiment is that better players committed more false recognitions
than weaker players, showing that in some circumstances expert per-
ception can lead to errors. In sum, these two experiments back up the
hypothesis that experts use anticipatory schemas in their perception:
rather than recalling a scene the way they saw it, experts tend to recall
it the way it normally unfolds in the near future.

PERCEPTION IS COGNITION
The importance and speed of perception is not limited to chess, but
has been documented in many other domains of expertise, such as
music, medicine, sports and driving. In all these domains, experts lit-
erally see a different problem situation and categorise it in a better way.
Rather than being innate, experts’ perception is the product of many
years of practice and study. One of de Groot’s major contributions is
to have shown that there is no clear boundary between perception
and cognition: in chess and in other domains, perception, memory
and problem-solving are tightly interconnected.
 2
CHUNKS!

De Groot uncovered some fascinating phenomena, but his work

was essentially descriptive and his theoretical explanations were not
compelling. The first theory to convincingly account for de Groot’s
results was proposed by Herbert Simon and William Chase in 1973 in
three classic papers, in what is known as the chunking theory. The theory
was primarily aimed at explaining two phenomena: chess experts’
remarkable memory and their ability to find good moves rapidly –
as de Groot put it, strong chess players automatically see the good
move. Chunking theory’s strength is to have proposed fairly detailed
mechanisms to explain these phenomena. The fact that Simon was
one of the founders of artificial intelligence and modern cognitive
psychology was not an incidental factor to the strength of chunk-
ing theory. Indeed, Simon had previously built several computational
models that captured some of the ideas he developed with Chase.
Chunking theory, as well as the empirical work that supported it,
motivated a considerable amount of research on expertise in the fol-
lowing twenty years or so.
Chunking theory assumes that chess players encode most of their
long-term memory knowledge as chunks – perceptual units that can
be treated as wholes. The first chunks are small, but then larger chunks
are incrementally built using these smaller chunks. In chess, chunks
14 C H U N K S !

consist, at the beginning, of individual pieces on a given square, and

then grow into groups of pieces. An analogy with reading will make
the process clear. At the beginning, a reader learns to recognise indi-
vidual letters, such as “t” and “h”. With practice, these letters form
chunks, such as “th”, and later “the”. The power of chunking is that
very large units indeed can be created by this mechanism. So, for
example, assuming much practice with reading, the following chunk
may be learnt: “To be or not to be, that is the question”. Such a chunk
is then a unit of both perception and meaning, and can be processed
as a whole. Figure 2.1 provides an example of the kinds of chunks
learnt by a weak chess player and a master.
In addition to mechanisms explaining how a network of chunks is
constructed, the theory made several assumptions about learning and
memory. It takes a fairly long time to learn a new chunk (8 seconds)
and to add information to a chunk already in long-term memory
(2 seconds). But once learnt, chunks can be retrieved rapidly, in a few
hundred milliseconds. Short-term memory capacity, which is limited
to seven items, is the same for experts and non-experts. Thus, the
main difference between weak and strong chess players is the number
and size of the chunks they have acquired. A final assumption is that
chunks can be linked to information. In chess, this information can

Figure 2.1 Chunks learnt by a weak player (left diagram) and a master (right
diagram).
 CHUNKS! 15

be a move or a sequence of moves, strategic ideas or tactical motifs.

For example, given a chunk encoding a pawn structure with a weak
square, the suggested action could be: “Place a knight on this square!”
In psychology and artificial intelligence such condition-action pairs
are known are productions.
Chunking theory explains the skill effect in recalling chess posi-
tions by assuming that strong players are more likely to recognise
chunks on a board position, since they have stored many more chunks
in long-term memory. Once a chunk is recognised, a pointer to it
is placed in short-term memory. Although only seven pointers can be
placed in short-memory, due to its limited capacity, pointers can be asso-
ciated to small, medium or large chunks. Since strong players
have acquired larger chunks than weaker players, they can encode
a position with a smaller number of units than weaker players, and
can memorise an entire position despite the limited capacity of
short-term memory. In sum, while novices perceive a position as a
collection of individual pieces, stronger chess players perceive it as
collection of familiar configurations.
In work with Kevin Gilmartin, Simon used computer simulations
and mathematical extrapolations to estimate that a chess master must
have learnt about 50,000 chunks. Simon and Chase note that this
number is roughly the same as the number of words that American
college students have in their vocabulary. Given the time needed
to learn these chunks, they estimated that it takes a minimum of
10 years, or 10,000 hours, to become a chess master.
Simon and Chase also addressed the question of how players use
mental imagery to visualise board positions and anticipate moves in
their mind’s eye. The central idea, already mentioned above, is that
potential moves are proposed by pattern recognition: patterns on
the board elicit chunks in long-term memory, which in turn suggest
possible actions. Chunks also provide information allowing players to
reconstruct groups of pieces in their mind’s eye. An important aspect
of the theory is that pattern recognition occurs not only when look-
ing at a physical board, but also when looking at a board position
imagined in the mind’s eye. In both cases, recognised chunks might
16 C H U N K S !

elicit information about what to do in a given situation, including

what kind of moves are likely to be good. Thus, this look-ahead search
consists of recognising chunks, using the information they provide to
update the position in the mind’s eye and repeating this process sev-
eral times. Because chunks provide useful information about moves,
plans, tactical motifs, etc., they enable a highly selective search. In
sum, chunking theory explains the speed at which expert players find
good moves by assuming that chunks allow them (a) to identify pat-
terns on the board; (b) to use these patterns to access useful informa-
tion, including potentially good moves; and (c) to repeatedly update,
after a move has been carried out, the board constructed in their
mind’s eye and explore the consequences of moves and sequences
of moves.
Simon and Chase’s theory was very ambitious, since it accounted
for data not only about memory, but also about problem-solving.
Moreover, it is not limited to chess expertise, but can also be applied
to other domains of expertise, and indeed to the study of cogni-
tion in general. This explains why it had considerable impact, not
only on expertise research, but also on cognitive psychology more
generally.

RECALL EXPERIMENT
Simon and Chase’s genius was not only to have proposed a power-
ful theory, but also to have supported it empirically with elegant
experiments. They focused on de Groot’s perception task – which
they considered as a recall task – providing both a replication and an
extension. A chessboard was shown for five seconds, after which play-
ers attempted to reconstruct it. There were two important differences
in comparison to de Groot’s experiment. First, Simon and Chase did
not ask participants to think aloud during the experiment. Second,
in addition to board positions taken from masters’ games, they used
random positions, where the pieces of a game position were haphaz-
ardly placed on the board. The results showed that there was a skill
effect with positions taken from games, but not random positions.
 CHUNKS! 17

An interesting feature of Simon and Chase’s study is that there

were only three participants. The master was Hans Berliner, a former
world chess champion by correspondence – a variant of chess where
moves are transmitted through post – who was a PhD student in
computer science at Carnegie Mellon University. The class A player
was William Chase himself. While Chase was aware of the hypotheses
being tested, the positions were not selected by him, for obvious rea-
sons. The beginner was Micheline Chi, a PhD student in psychology,
who later married Chase. The small sample was compensated by very
detailed analyses, which tested very specific theoretical questions. This
study is a nice example of the fact that experiments with very small
sample sizes can lead to powerful results, which in this case started a
very influential tradition of research.

COPY TASK
In comparison to the original study, another important addition was
a copy task, which provided critical information about the structure of
chunks. In this task, players could still see the stimulus board when
the position was reconstructed on a second board. So, typically, play-
ers would glance at the stimulus board, direct their attention to the
reconstruction board and place a few pieces, glance again at the stim-
ulus board, and place yet more pieces. Simon and Chase used the fact
that the two boards could not be fixated at the same time to provide
an empirical operationalisation of a chunk: pieces replaced together
after a glance at the stimulus board belong to the same chunk. Thus,
complementing the first definition of a chunk as a group of pieces,
we now have here a second definition.
Simon and Chase also studied the latencies between the placements
of two pieces. They reasoned that, if two pieces belong to the same
chunk, they should be replaced rapidly together. By contrast, if two
pieces belong to two different chunks, the latency between them
should be longer. As predicted, most of the pieces within a chunk – as
defined by whether players glanced or not at the stimulus board –
were replaced in less than 2 seconds, while the pieces that belonged
18 C H U N K S !

to two different chunks were replaced in more than 2 seconds. In

addition, the distributions of placement times between the recall and
the copy task were essentially the same. Thus, one can assume that
the same process occurs in the recall and the copy task. Here, we have
a third definition of a chunk: the 2-second threshold between the
placement of two pieces.
Finally, Simon and Chase used chess semantics to define a chunk.
They counted the number of chess relations shared by two pieces
placed in succession. There were five relations: colour, defence, attack,
proximity and kind. Simon and Chase found that two pieces placed
in succession shared more chess relations with each other when
they were part of a chunk than when they belonged to two different
chunks. The beauty of Simon and Chase’s research is that the three
empirical definitions of a chunk – glances at the stimulus board,
latencies and number of relations – provided very consistent results.

DIRECT SUPPORT FOR CHUNKING HYPOTHESIS

As discussed below, Simon and Chase’s study has been replicated with
larger samples and a tighter methodology. In addition, the expertise
effect in recalling chess material is robust and has been obtained
even when the experimental method is changed in various ways. For
example, the effect remains when the presentation times are varied
from 1 second to 60 seconds, when positions come from openings,
middlegames and endgames or when positions are presented on dia-
grams where chess pieces are replaced by letters. The skill effect has
also been found in nearly all domains of expertise, including games,
sports, science and the arts.
The chunking hypothesis has also been directly confirmed in sev-
eral experiments. The earliest support, and one of the most convinc-
ing, was provided by Neil Charness in his PhD thesis. He dictated
positions using the notation used by chess players, at the pace of just
above 2 seconds per piece. The pieces were grouped during dicta-
tion, using three different orders. In the first condition, Charness
used the kind of chunks defined by Simon and Chase; in the second
 CHUNKS! 19

condition, the pieces were dictated file by file; in the final condition,
the pieces were presented randomly. In line with chunking theory,
players recalled the positions best when the pieces were dictated using
chunks. Random order yielded the worst recall. In another experi-
ment, Charness presented the pieces visually, with the same results.
These experiments are important, because the variable of interest –
whether pieces belonged to the same chunk or not – was controlled
by the experimenter.

PROBLEMS WITH THE THEORY

Unfortunately, several experimental findings turned out to be incon-
sistent with chunking theory. In fact, there were two anomalies in
Simon and Chase’s own results. The first was that the largest chunks
replaced by their master were not that large: a maximum of 5–6 pieces.
The second anomaly was that the master and class A player employed
more chunks than the novice. In all cases, the number of chunks did
not exceed the postulated capacity of short-term memory (7 ± 2),
but this result could be an indication that stronger players’ short-term
memory has a larger capacity, contrary to the prediction of the theory.
Although they raised a lot of theoretical debate and led to numer-
ous experiments, these two anomalies were in reality an artefact of
the way Simon and Chase presented the positions, using a standard
chessboard and pieces. The issue here is that the hand can only hold a
limited number of pieces; thus, even if players recognise large chunks,
these are going to be split in smaller chunks. This explanation was
confirmed by two experiments I carried out, the first with Herbert
Simon himself, and the second with Gary Clarkson, who was an
undergraduate student at the University of Nottingham. In the first
experiment, a computer display was used to present the positions and
record the participants’ reconstructions. We also had a much larger
sample than in the original study. This time, masters replaced very
large chunks, containing up to 19 pieces – about three times larger
than what Simon and Chase had found. Also, the number of chunks
replaced did not differ between players of different skill levels.
20 C H U N K S !

In the second experiment, we found direct support for the hypoth-

esis that the capacity of the hand confounded Simon and Chase’s
results. Players performed the copy and the recall task, both with
a physical board and a computer display. Just like with Simon and
Chase’s study, the physical display led to relatively small chunks. And
just like with the replication I did with Simon, the computer display
led to fairly large chunks. It should be noted that the presence of large
chunks is consistent with the way chess masters describe their games,
as was documented in the verbal protocols collected by de Groot.
Masters employ high-level concepts, in some cases characterising the
entire board. For example: “a queen’s gambit defence, with White
carrying out a minority attack and Black having a passive position”.
While the anomalies in Simon and Chase’s data were artefacts,
other findings posed genuine challenges for chunking theory. A first
recalcitrant result was uncovered by Neil Charness, again in his PhD
research. As we saw earlier, two of the assumptions of chunking the-
ory are that short-term memory has limited capacity and chunks
are encoded in long-term memory relatively slowly. When these two
assumptions are combined, a strong prediction is that short-term
memory should be highly sensitive to interferences. Thus, interpolat-
ing a task between the presentation of a chessboard and its recon-
struction should lead to a drastic loss of performance. This is because
the pointers to long-term memory are erased by the second task;
since learning new information is slow, there is not enough time to
encode the board in long-term memory. In fact, this is what happens
with the recall of words when interpolated tasks are used. However,
contrary to the prediction of chunking theory, Charness found that
the interfering task decreased performance only by about 10%, which
is relatively small. Even using chess tasks as interference, such as find-
ing the winning move in a second position, did not affect the recall
of the first position much.
One way of creating interference is to ask players to recall two,
three or more positions. The positions that come later in the sequence
can be considered as interfering tasks for the positions that came ear-
lier (retro-active interference). Indeed, retaining the earlier positions
 CHUNKS! 21

also interferes with the memory of the positions that come after (pro-
active interference). Again, chunking theory predicts that recall of all
positions should become increasingly poorer as the number of boards
increases. Several authors carried out the experiment, and the results
showed that, while the task was too hard for amateurs, masters can
do it relatively well, even though there seems to be a barrier after
five boards.

A REVISION OF CHUNKING THEORY

Chunking theory did a good job at explaining some of the empirical
data, but failed to account for the lack of interference effects. Could it
be modified so that its weaknesses are corrected whilst its strengths
are maintained? This was the aim of template theory, which I devel-
oped with Simon. The key idea was to offer mechanisms explaining
how high-level memory structures (schemas) are unconsciously cre-
ated from simpler memory structures (chunks). There is substan-
tial empirical evidence that people use schemas: these are structures
where some parts are invariable, while some parts are variable. A good
example is the schema of a house. A house normally has a floor, a
ceiling and a roof; this is the fixed part of the schema. In addition,
some components might or might not be present (such as a garage,
an en-suite and a basement), and the number of some components
might vary (such as the number of doors and windows); this is the
variable part of the schema.
Template theory keeps the idea of chunks but also proposes that
templates – which are a kind of schema – are created when some
chunks are used frequently. The core of the template, which is a
chunk, consists of stable information. The slots consist of informa-
tion that is variable. For example, in chess, a slot could be created for
some pieces that are important, given the presence of the chunk in
the core, but that can be located on different squares. Similarly, a slot
could be created for an important square that could be occupied by
different pieces. The theory assumes that information can be encoded
rapidly in the slots, in about 250 milliseconds. Thus, the presence of
22 C H U N K S !

templates accounts for the results of the interference and multiple-

board experiments not only by providing fairly large memory struc-
tures, but also by assuming rapid encoding into long-term memory,
thus side-stepping the limited capacity of short-term memory. Impor-
tantly, templates can be used to store domain specific information
only. Thus, chess experts can use them to memorise chess positions,
but not lists of words. Just like chunks, templates also provide use-
ful information for decision-making, such as plausible moves and
standard plans. In particular, they allow search to be carried out in a
more abstract space than considering concrete moves. As documented
in verbal protocols, strong players often carry out search using key
points of a game. For example, a standard plan for Black in the Benoni
defence is (a) to expand on the queen’s side, (b) to exchange pieces
and (c) to win the endgame due to the pawn majority on the queen’s
side. The claim is that templates offer the type of information neces-
sary for carrying out this kind of abstract search.
Thus, template theory postulates more learning mechanisms than
chunking theory. In addition to chunks and productions, already pres-
ent in the original theory, template theory postulates that templates
must be acquired, and so must links connecting chunks and templates
together if they are similar enough. Constructing such a network,
which is more complex than that proposed by chunking theory,
explains why it takes several years of study and practice to become
an expert in chess and other complex domains.
The great strength of template theory is that it is implemented as
a computer programme (CHREST: Chunk Hierarchies and REtrieval
STructures), which guarantees that all the postulated mechanisms are
specified in detail (if they are left unspecified, then the programme
will not run). All the learning mechanisms mentioned in the previ-
ous section are implemented, as well as mechanisms for managing
short-term memory and focusing attention through eye movements.
Importantly, all the learning is done automatically, simulating the
unconscious acquisition and use of knowledge by human players.
CHREST has simulated several of the empirical data I have reviewed
in this and the preceding chapter. From beginners to grandmasters, it
 CHUNKS! 23

replicates the percentage of pieces correctly recalled, the way pieces

are chunked during reconstruction and the type of errors made. It
also simulates how different types of positions (e.g. positions taken
from masters’ games, positions modified by mirror-image and vari-
ous types of randomised positions) affect recall. It also accounts for
how manipulating presentation times from 1 second to 60 seconds
affects recall, thus providing support for the validity of CHREST’s time
parameters, such as the time to create a new chunk (8 seconds), and
the time to add information to a template slot (250 milliseconds).
In the previous chapter, I described the clear differences between
novices and masters’ eye movements when they look for 5 seconds at
a new position with the aim of memorising it. CHREST can replicate
these differences very well. Just like human players, the simulated
masters have shorter fixation times, show less variability in the dura-
tion of their fixations, tend to fixate important squares and fixate
more squares as well as more important squares. These results are
partly due to the fact that, while the novice version relies exclusively
on simple but relatively slow rules of thumb (e.g. fixate the centre)
to direct eye movements, the master version can also use the infor-
mation contained in chunks, which enables faster fixations as they
are automatic. As more chunks have been learnt, more fixations are
directed by the structure of chunks in long-term memory.
CHREST has also accounted for many empirical results beyond
chess, such as aspects of expertise in the African game of awele and
the East Asian game of Go. Beyond board games, it has simulated
phenomena related to concept formation, implicit learning and even
acquisition of first language, which can be considered as a kind of
expertise.

RECALL OF RANDOM POSITIONS  AN

UNEXPECTED BUT CORRECT PREDICTION
An interesting contribution of CHREST is to have made new and
counterintuitive predictions that were later supported by the data.
Remember that random positions are created by taking the pieces of
24 C H U N K S !

Figure 2.2 Examples of a position taken from a game (left diagram) and a random
position (right diagram).

a game position, and randomly placing them on the board (see Fig-
ure 2.2 as an example). Remember also that, with random positions,
Simon and Chase did not find any difference in performance between
their master, class A player and novice. This pattern of results – vast
skill differences with game positions, but no differences with random
positions – was both spectacular and intuitively satisfying: experts can
use their knowledge with structured material but not with unstruc-
tured material. It thus found its place in numerous textbooks and
popular-science books. The problem is that the second part of the
argument (no skill effect with random positions) is not true.
When I was developing CHREST in collaboration with Simon,
the programme consistently and stubbornly predicted that masters
should remember random positions better than weaker players, even
though the absolute number of pieces remembered should be much
smaller than with game positions. In one of the rare cases where his
scientific intuition let him down, Simon did not believe the simula-
tions. He thought there were mistakes in the computer programme.
There were small mistakes, for sure, but once corrected, the behav-
iour of the programme was the same: it predicted a skill effect with
random positions. The mechanism behind this behaviour was actu-
ally simple and a direct consequence of chunking theory. By chance,
there are patterns even in random positions, albeit not many of them.
 CHUNKS! 25

Masters, who have acquired more chunks than weaker players, are
more likely to recognise one of these patterns by chance and thus
access a chunk in long-term memory. Thus, they should show a small
superiority over weaker players.
We combed the literature to find all studies that had used random
positions in experiments on chess memory. These positions were typi-
cally used as a control task to ascertain that better players do not have a
better memory in general. When put together, the results indicated that
there is a skill effect even with random positions, even though it was
smaller than with game positions. On average, masters recalled about
20 pieces with game positions and five pieces with random positions,
while weak amateur (below class B players) recalled five pieces and
2.5 pieces, respectively. In fact, a skill effect was found in 12 out of 13
studies, the exception being Simon and Chase’s study! The skill effect
did not reach statistical significance in most studies, due to their small
sample sizes. In the end, Simon conceded that he should have believed
the computer simulations rather than his own intuition.
This result is important theoretically. Not only is it a direct predic-
tion of chunking theory, but it is actually difficult to explain for most
theories of expertise – for example, theories that assume that experts’
knowledge is primarily coded by relatively high-level memory struc-
tures such as schemas. The reason is that small memory structures
such as chunks seem necessary for explaining this result.
There is very recent twist in the story. In a meta-analysis of the
studies having used random material not only with chess, but with
other domains of expertise as well, my PhD student Giovanni Sala
found that this skill effect generalises to most domains of exper-
tise, thus providing additional support for the existence of chunking
mechanisms in human cognition.

OTHER TYPES OF KNOWLEDGE

Chunks (perceptual knowledge) and potential moves (procedural
knowledge) are important kinds of knowledge in chess, but this is
not the full picture. Strong chess players possess additional types of
26 C H U N K S !

knowledge. For example, they have a large stock of declarative knowl-

edge. This includes knowledge of openings (it has been estimated that
masters have memorised at least 100,000 opening moves), tactical
motifs (“if a piece fulfils several functions and is overloaded, try to take
advantage of this”), strategic principles (“two bishops are usually bet-
ter than two knights”) and endgame principles (“passed pawns should
usually be advanced”). This also includes an arsenal of techniques and
methods such as “if you’re one pawn up, try to exchange pieces but
not pawns”. Note that this declarative knowledge becomes increasingly
refined as expertise grows. For example, amateurs usually know that
“occupying an open file with rooks is advantageous”. Better players,
perhaps at master level, have learnt the maxim that “occupying an open
file with rooks is advantageous, unless the opponent controls all entry
points with minor pieces”.Yet a stronger player would know that “occu-
pying an open file with rooks is advantageous, if, in the case where the
opponent controls all entry points with minor pieces, it is possible to
penetrate the opponent’s position with an exchange sacrifice”. Some
of this knowledge is clearly declarative, but some of it is implicit and
encoded in thousands of pairs of chunks/actions. Whatever the type of
knowledge, there is the question as to when it is useful and should be
acted upon, and when it should better be left alone. It is not uncom-
mon that a small change in the position leads to tactical opportunities
that invalidate even the best strategic principles. The answer is that look-
ahead search should be carried out to detect such instances.

BLINDFOLD CHESS
There is a variant of chess that supremely taxes memory: blindfold
chess. With blindfold chess, one plays without seeing the board,
unlike the opponent who, typically, can see it. Players normally com-
municate using the algebraic notation (e.g. 1. d4 Nf6 2. c4 g6, etc.),
which is the standard way of recording moves. As if playing one
blindfold game was not difficult enough, it is also possible to play
several blindfold games at the same time. Whilst most players above
candidate master level can play one blindfold game without too
 CHUNKS! 27

much difficulty, playing several games is a special skill that demands

dedicated training. A major difficulty is to keep boards as distinct as
possible, which requires a smart choice of openings, for example
alternating quiet variations with wild gambits. In addition, players
use several mnemonic tricks, such as planning in advance the kind of
opening they play on a given board, and making associations between
the face or the voice of the opponents and the game itself.
The current world record is a staggering 48 games, established in
2016 in Las Vegas by Uzbek grandmaster Timur Gareyev, who beat the
record held by German master Marc Lang by two games. During the
entire exhibition, Gareyev wore a blindfold mask and pedalled on an
exercise bike, which he found useful to keep relaxed and focused. After
19 hours and 9 minutes – which included a 30-minute break due
to a fire alarm! – and about 1,400 moves, he obtained a result of 35
wins, 7 draws and 6 losses, thus scoring 80.2% of the points. (He also
rode more than 80 km!) Gareyev prepared for this event for several
months. He used a variety of mnemonics including the “memory pal-
ace”, where visual and conceptual associations are made with a list of
pre-learnt locations, and the construction of connections using charac-
teristics of the game, the board number and the name of the player. In
an interview, he gave the examples of playing a risky opening on board
13 to try his luck and of associating the capture of a piece with a burst.
Starting in 1894 with Alfred Binet, the inventor of intelligence
tests, there have been many informal accounts of blindfold chess
in the literature. Reuben Fine, a world-class grandmaster who later
became a psychoanalyst (see chapters 5 and 10) provided an inter-
esting introspective account in a paper written in 1965. According
to him, chess knowledge is essential. In particular, strong players use
hierarchical “spatio-temporal Gestalts” that make it possible to under-
stand the position as a whole. Although Fine was one of the few play-
ers who also stressed the capacity to visualise the board clearly, he
also mentioned that pieces are perceived dynamically, not just as static
objects. Additionally, he stressed the importance of using key state-
ments for summing up positions (e.g. “a Sicilian defence, Taimanov
variation, badly played by Black who has very weak dark squares”).
28 C H U N K S !

A fair amount of research has been performed on blindfold chess.

For example, Finnish psychologist Pertti Saariluoma carried out sev-
eral fascinating experiments where he combined blindfold play with
manipulations of memory load. Moves from a game were dictated
at the pace of one per second. After 15 and 25 moves, players had
to indicate the location of each piece. To spice things up, players had to
perform an interfering task during the presentation of the game. This
task was either verbal (to repeat the syllable “tik”) or visuo-spatial (to
form a mental image of three capital letters and mentally walk along
the sides of these letters, saying whether the turns are left or right).
There was also a control condition without any interfering task. The
results showed that the visuo-spatial, but not the verbal task, had a
negative effect on the performance in blindfold chess. In another
experiment, the interfering tasks were given after the sequence of
moves had been dictated. There was no loss in performance in either
condition. Based on these and other experiments, Saariluoma con-
cluded that blindfold chess requires visuo-spatial working memory
but not verbal working memory. He also inferred that, more than
differences in imagery ability, knowledge is the essential ingredient
of blindfold chess, and specifically knowledge coded as chunks linked
to possible actions.
Blindfold chess raises an interesting paradox. If strong players
are able to play a game without seeing the board – as grandmasters
clearly do – why is it that they do not do it in normal games? Surely,
staring at the board in front of them creates serious interferences with
the positions that they imagine in their mind’s eye during look-ahead
search. In fact, some players such as Ukrainian grandmaster Vassily
Ivanchuk occasionally do so. The reason that most players do not do
it can probably be explained by the fact that the physical chessboard
provides visuo-spatial information that supplies memory support for
the location of pieces. Still, as we shall see in Chapter 7, there are
well-documented cases were players made errors in their calculations
because they thought a piece was still on the square it was located
on the external board, and not on the square it had moved to during
look-ahead search.
 3
THE BEST MOVE

The previous two chapters have provided considerable evidence for

the role of perception and knowledge in chess. But obviously, chess
players must be doing something else than just recognising good
moves automatically. Else, why do they not play immediately, rather
than pondering about their next move, sometimes for very long peri-
ods of time? For example, in the 14th game of the world champion-
ship match in Merano in 1981, Korchnoi thought for 1 hour and 20
minutes on his 13th move before moving his bishop one square up
in reply to Karpov’s knight’s move.
The beauty of chess is that the number of possibilities grows
extremely rapidly. There are about 35 legal moves in a middle game
position, on average. Anticipating all possibilities two moves ahead
requires considering 35 × 35 moves = 352 = 1,225 moves. The
number of legal continuations six moves deep already reaches 356 =
1.83 billion. Obviously, numbers keep increasing exponentially for
greater depths. No player can carry out a systematic search – not even
Korchnoi in 1 hour and 20 minutes! – except for trivial positions.
As pointed out by de Groot, look-ahead search must be highly
selective and informed by chess knowledge. In his experiment, even
world-class players rarely considered more than 100 possible con-
tinuations before making their move. But how do players carry out
30 T H E B E S T M O V E

their search? Do they sometimes search very deeply? What kind of

knowledge do they use? This chapter will answer these questions.

DE GROOT’S SURPRISING RESULTS

Again, the starting point is de Groot’s PhD thesis. It aimed to under-
stand how chess players think, and in particular what are the dif-
ferences between strong players – masters and grandmasters – and
weaker players. The chess literature offered two contradictory views.
The majority of the authors argued that the ability of looking ahead
was central to chess expertise, often referring to the deep combi-
nations dreamt up by world champion Alexander Alekhine. Others
argued that selectivity was key. When asked how many moves he nor-
mally saw ahead, grandmaster Richard Réti, one of the best players in
the 1920s, is said to have replied: “Only one move, but the best one!”
As we have seen in Chapter 1, de Groot used a simple experimental
procedure. He showed participants a chess position, and asked them
to select their next move, as if they were playing one of their own
games. This is a very common task for chess players, who also do this
when they practice. In addition, de Groot asked them to think aloud.
Again, this is not uncommon for chess players, who are used to talk-
ing together about their games and analysing moves. As there were
no video recorders or smart phones at the time, de Groot would jot
down on paper what the players said, and then clean up his notes.
His sample was extraordinary, with six world-class players, including
world champions Max Euwe and Alexander Alekhine. In addition,
the sample included four Dutch masters, five candidate masters, two
female Dutch champions and five weaker players.

NO SKILL DIFFERENCES IN THE STATISTICS

OF SEARCH
As expected, grandmasters and masters selected better moves than
weaker players. To understand how this difference came about, de
Groot analysed the verbal protocols produced by the players both
 THE BEST MOVE 31

quantitatively and qualitatively. Nowadays, only the quantitative anal-

yses are remembered in the scientific literature, which is a pity, as the
more interpretative analyses tell us a lot about the cognitive processes
underpinning chess players’ thinking.
A protocol looks like this, omitting the chess notation:

Great position. I have a strong attack. I have three possible moves:

I can take her rook, retreat my rook or attack her queen with my
knight. If I take her rook, she takes back, I can either give check
with my bishop or attack her queen with my knight. In both
cases, I’m winning. But she can also simply move her king, this
is very annoying. So, in my first move, I can also retreat my rook.
I don’t like this. I can also attack her queen with my knight. She
must move her queen here and then I give check with my bishop.
It’s winning. Let’s calculate again. I attack her queen, she goes
here, and I give check. OK, I play this move.

A player’s thoughts can be displayed by a search tree (see Fig-

ure 3.1). Each node represents a position, and the moves are indi-
cated on the branches. This representation makes it easy to compute

Take Attack
Retreat White
rook queen
rook

Take Move Move

Black
back king queen

Give Attack Give White

check queen check

Figure 3.1 Example of a search tree.

32 T H E B E S T M O V E

measures such as the number of different moves considered (nine in

the example) and the maximal depth of search (three moves).
The quantitative analyses focused on the differences between
grandmasters and candidate masters. De Groot devised about 15 mea-
sures, including the time to make a decision, the number of different
first moves considered, the number of different positions explored
during search and the average depth of search. To de Groot’s surprise,
grandmasters and candidate masters did not differ with respect to
the majority of these variables. In particular, depth of search was
about the same between the two groups. In addition, all players were
highly selective; out of about 30 possible first moves for White in
the problem position, players considered only about three or four
of them. Even when differences were found, such as with the rate
of generating moves, they were small and insufficient to explain the
large disparity in the quality of moves chosen.

PROGRESSIVE DEEPENING
The qualitative analyses describe, in great detail, the way the deci-
sion processes are structured. Most of de Groot’s book is devoted to
them. An important outcome was that all players were investigating
the same base move (one of the first moves in the problem situation)
several times – a process that de Groot called progressive deepening. The
reinvestigation could be immediate, or separated by the analysis of
another base move. In either case, reinvestigating a base move makes
it possible for the player to study it with more precision, by increas-
ing the number of moves anticipated or by improving the evaluation
of the terminal position.
At first blush, this process seems repetitive and inefficient, since
one might think that resources could be better used by analysing
different base moves and new sequences of moves. However, progres-
sive deepening is actually quite adaptive. First, it helps alleviate the
restrictions imposed by the limited capacity of short-term memory.
Revisiting the same sequence of moves several times increases the
likelihood that it is stored in long-term memory, thus making analysis
 THE BEST MOVE 33

easier as more cognitive resources can be expended on it. Second, it

makes it possible for players to use information that has been gleaned
at different places of the search tree.
Finally, progressive deepening reflects the cyclic organisation of
chess players’ thought, which consists of several observation-test-
evaluation phases. Anticipating ideas later developed by Newell and
Simon during the cognitive revolution in psychology, de Groot pro-
posed that the thoughts of a chess player can be described as complex
hierarchies of problems and sub-problems. To deal with sub-problems,
larger cycles may include smaller cycles. For de Groot, progressive
deepening is not unique to chess, but is a fundamental characteristic
of the way humans tackle complex problems and make decisions. As
a paradigmatic example of this, de Groot mentions scientific research,
where a researcher explores one possible solution, then examines
another one, then goes back to the first solution and so on. In science,
this process can take minutes, hours, days or even years.

THE PHASES OF CHESS THINKING

According to de Groot, chess players’ thinking is typically divided into
four phases: orientation, exploration, investigation and proof. During
the orientation phase, players gather relevant information and provide
a first preliminary evaluation of the position. This is followed by the
exploration phase, where players analyse sample variations and progres-
sively reduce the number of critical moves to two. These two moves
are carefully examined during the investigation phase, where search is
carried out at a deeper level than during the exploration phase. Dur-
ing this phase, players also try to find support for their favourite move.
Interestingly, most of the arguments deployed by chess players are
really an attempt to convince themselves that one of the two moves
they had selected is better than the other. The final phase, called the
proof phase, sees players recapitulate the conclusions obtained in the
previous phase and double-check that the argumentation is correct.
As we have seen in the previous chapter, players rely on a great
deal of knowledge, which includes strategic goals and plans, tactical
34 T H E B E S T M O V E

motifs, specific methods for converting a material advantage and so

on. There are vast skill and individual differences in the extent and use
of this knowledge, which is domain specific (i.e. limited to chess).

MACROSTRUCTURE OF SEARCH: THERE

ARE SKILL DIFFERENCES!
While most of de Groot’s conclusions were insightful, there was
ample space for further research to expand our understanding of
chess players’ thinking. Remember that de Groot found only very few
skill differences with respect to the quantitative variables describing
search. However, one should not forget that he compared grandmas-
ters with candidate masters, who are fairly experienced players. In
fact, when later researchers included weaker players in their sample, a
number of differences became apparent. At the extreme, most begin-
ners do not have the ability of anticipating any moves in their mind’s
eye, and thus the difference between grandmasters and novices is
considerable. Differences remain when more experienced amateurs
are considered, although these differences are not particularly large.
For example, depth of search increases rapidly in the early stages of
expertise, but improvement then becomes slower and slower; this
kind of improvement, called the “power law of learning”, is common
in the literature of skill acquisition. With a power law, it is hard to
detect skill differences with highly skilled players, since differences
rapidly become minuscule. The data show that, on average, class D
players search 2.3 moves deep, candidate masters 4.8 moves deep,
and top-level grandmasters 5.3 moves deep. The last two data points
are from de Groot’s seminal study, and the difference is not large
enough to detect statistically with a small sample.
Most statistics in de Groot’s study were collected with a relatively
simple position. When unfamiliar, complex and tactical positions
are used, the skill differences in search increase considerably. For
example, in a study carried out with Guillermo Campitelli, we chose
“crazy” positions where the best moves were very difficult to find.
The grandmaster had an average search of 13.8 moves, with some
 THE BEST MOVE 35

lines being searched 25 moves deep. The corresponding numbers for

the class B player were 2.8 moves and 10.5 moves – a huge difference!
Thus, strong players can adapt their search strategy to different kinds
of positions, while this is not possible for weaker players. This seems
to suggest that experts are more flexible with their use of search
strategies than non-experts.
Another interesting result is that there are skill differences with
respect to progressive deepening. A branch in the search tree can
be re-examined in two different ways: either directly after (immediate
reinvestigation), or after having examined one or several other branches
(non-immediate reinvestigation). Re-analysing data I had collected as an
undergraduate student, I found that stronger players tend to use more
immediate reinvestigations and fewer non-immediate reinvestigations
than weaker players. Strong players seem to use a win-stay and lose-
shift strategy, to use terms from game theory. If the evaluation of a
sequence of moves is positive, they search it again at a deeper level,
verifying that the opponent does not have any moves that refute it.
If the evaluation is negative, they consider the line as unfavourable
and examine other moves, widening their search. In most positions,
strong players rapidly identify promising moves by pattern recogni-
tion (see Chapter 2) and thus are likely to stay with them rather than
considering other moves.

JUDGEMENT AND PLANNING

Judgement and Planning in Chess, written in 1952, is a famous book for
intermediate-level players, in which world champion Max Euwe pro-
vides advice about topics such as weak pawns, strong squares and the
pawn majority on the queenside. Judgement and planning are indeed
two central concepts in chess, and Euwe’s book is only one of the
many treatises in the practical chess literature advising players how
to excel in such skills. Judgement concerns the evaluation of the posi-
tion. Do I have the advantage? What are my strengths and weaknesses,
and those of my opponent? In a very old paper (1907!), Cleveland
reports the results of a questionnaire that clearly show that stronger
36 T H E B E S T M O V E

players evaluate positions better. More surprising is the observation

made by de Groot in his thesis that evaluations typically concern
only one aspect of the position, and more rarely two or three. For
example: “Black is better because White’s square f3 is very weak”.
There are actually also many instances in his protocols where players
provide only broad evaluations such as “Black is better”, without any
supporting reason.
The way humans evaluate chess positions is a very far cry from
the way computers do it, putting together dozens of factors such as
control of the centre, piece mobility, material balance and king safety.
It is as if players first evaluate the position unconsciously and auto-
matically, presumably by pattern recognition, and then come up with
one or two features to justify their evaluation. Obviously, this is not
consistent with full rationality, according to which all the features of
the position should be combined mathematically in the computation
of the evaluation function, just as is done by computers.
In the late 1970s, Dennis Holding carried out several experi-
ments in order to understand how chess players judge positions.
In one experiment, he estimated the quality of players’ evaluations
as a function of depth of search. A computer programme provided
an estimation of the correct evaluation. In general, stronger players
provided better evaluations than weaker players, and the best evalu-
ations were for the positions as they were on the external board. In
addition, as the positions imagined during look-ahead search went
deeper in the search tree, and thus farther from the problem situa-
tion, the quality of the evaluations went down. Another interesting
result was that better players were more discriminative than weaker
players in their judgements, giving high scores to positions where
they had a clear advantage, and low scores to positions where the
opponent had a clear advantage. By contrast, the weaker players
tended to think that both sides had about equal chances, even when
this was not the case.
Planning is the setting of long-term goals and the way to achieve
them. Unlike the type of search I have discussed so far, which is
typically move by move, planning omits many details and often does
 THE BEST MOVE 37

not include specific moves. We know from anecdotal evidence that

strong players are good at finding a correct plan. Indeed, as already
noted by de Groot, strong players often perceive the correct plan in a
few seconds, just by recognising the type of position they are facing
and retrieving similar situations from their long-term memory. This
is often an instance of case-based reasoning: this position reminds
me of the game X vs. Y, thus the plan should be first A, then B and
finally C. At least in typical positions, plans are retrieved without any
look-ahead search or even any explicit evaluation of the position. Of
course, when the tactics turn out unfavourably, there might be a need
to find a new plan. For example, from the original plan of attacking
the opponent’s king, one has to satisfy oneself with the less glorious
plan of defending a weak pawn. Somewhat surprisingly, there is not
much scientific research on planning in chess.

SCHEMAS AND MIND SETS

As seen in the previous chapter, de Groot highlighted the role of
schematic knowledge in perception and problem-solving. Much of
this knowledge is reproductive, which means that players, when pos-
sible, tend to choose stereotypical but efficient solutions over creative
solutions. This knowledge enables masters to play good chess without
much search or thinking, simply by applying known methods and
principles. Experiments carried out by Pertti Saariluoma as well as
Merim Bilalić and colleagues have buttressed this hypothesis. When
presented with problems with stereotypical solutions, players tend to
find these solutions, even when shorter but more original solutions
are present. Does this mean that players become more rigid as their
expertise increases?
In a series of clever experiments, Bilalić and colleagues showed
that this is not the case. They induced mind-set effects (also known
as Einstellung effects) by creating problems that had two solutions: a
familiar but longer solution and an unfamiliar but shorter solution.
Not only did players spot the familiar solution rapidly, but they could
not find the shorter solution when requested to do so. It is not that
38 T H E B E S T M O V E

the shorter solution was particularly hard. It was found rapidly by

a control group who saw slightly modified positions so that only
the shorter solution was possible. The mind-set effects were present
even with candidate masters and masters. However, the crucial result
was that these effects became weaker as the strength of the players
increased. Thus, the more expert players were less rigid than the less
expert players.
Mind-set effects pervade human cognition, as for example can
be seen with racist and sexist stereotypes in social psychology and
confirmation bias in cognitive psychology. Be it in politics, science
or relationships, people tend to select information that supports their
views and ignore information that challenges them. Could chess shed
light on the mechanisms that underpin these mind-set effects? Bilalić
and colleagues recorded the eye movements of candidate masters
confronted with the type of two-solution problems discussed above.
As in previous experiments, players found the stereotypical solution
rapidly and failed to find the short solution. Here comes the fascinat-
ing part of the results: all players said they looked at other moves and
tried hard to find the shorter solution. However, their eye movements
told a different story: players kept looking at the squares critical for
the stereotypical solution, even though they thought that they were
looking at other parts of the board. It is as if, once a schema linked to
the familiar solution had been activated by perceptual patterns on the
board, players could not inhibit it. Beyond chess, these results raise
profound and troubling questions about introspection, consciousness
and free will.

PLAYING AGAINST KASPAROV

This chapter has shown that there is considerable evidence for the
importance of search in chess decision-making. Similarly, the previ-
ous two chapters have documented the great significance of pattern
recognition. There has been much debate in psychology about the
respective roles of these two mechanisms. Whilst de Groot and Simon
emphasised the roles of perception and knowledge, they also agreed
 THE BEST MOVE 39

that look-ahead search was important. However, they both stressed

that search was highly selective (rarely more than 100 positions),
precisely because of the knowledge that players can bring to bear. In
addition, Simon repeatedly pointed out that search must be limited
because of the constraints imposed by the limited capacity of short-
term memory and the bottleneck of attention.
There is also substantial empirical evidence for pattern recogni-
tion. To begin with, support is offered by strong players’ ability to
zoom in rapidly on promising moves, as already documented by
de Groot and later corroborated by more experimental research.
Then, there is the fact that grandmasters maintain a fairly high
quality of play when they compete in speed chess tournaments,
where they have about 5 seconds per move – a drastic reduction
in comparison to standard games (2–3 minutes per move). Finally,
compelling evidence is provided by simultaneous displays, where
one player (typically a master or a grandmaster) plays against 20
or 30 amateurs. These exhibitions are spectacular, not least because
the master typically wins most games, in spite of spending only a
few seconds on each move. In fact, simultaneous games are tough
not because of the difficulty of the games – the difference in
expertise is such that playing normal moves and waiting for the
opponent to make a mistake is normally sufficient for winning a
game – but because the master must walk several kilometres in
the process!
Perhaps the most arresting example of simultaneous chess was
offered by world champion Garry Kasparov, who played against
entire national teams! For example, in 1992 he beat the German
team, which consisted of four seasoned grandmasters, by the score
of 3 to 1. Simon and I used the results of seven such matches and
two matches against the team of Hamburg to argue that knowledge,
accessed by pattern recognition, plays a larger part in chess expertise
than does anticipating moves ahead. Unlike standard simultaneous
displays, where the master always has White and thus enjoys a small
starting advantage in all the games, in these matches Kasparov had
White in half of the games and Black in the other half. Out of the
40 T H E B E S T M O V E

nine matches he had played at the time, Kasparov won eight of them;
he lost only his first match, presumably because he was new to this
type of event. His median performance was 2646 Elo, which was
the strength of a top grandmaster at the time and would still have
placed him in the best six players in the world. In fact, Kasparov’s
performance in these matches was typically less than 100 points
below the level displayed in normal tournament play. For example,
his rating in July 1987 was 2735.
The teams consisted of four to eight strong masters and grandmas-
ters, and thus Kasparov’s thinking time was reduced proportionally. As
search is carried out serially and relatively slowly (perhaps 10 posi-
tions per minute, based on de Groot’s data), reducing thinking time
should affect the number of positions examined and consequently
decrease performance. The loss should not be as large if pattern rec-
ognition played a larger role, because this process is assumed to occur
rapidly. That is what was observed in the actual results. In line with
this analysis, there was no correlation between Kasparov’s perfor-
mance and the number of opponents he was facing, which directly
affected the time available for his thinking.
While I still believe that this analysis is correct, two factors should
be taken into account. The first is that Kasparov did carry out a fair
amount of search, as is obvious from looking at the games he played
in these matches. The opposition essentially consisted of professional
players, including many grandmasters; therefore, he could not just
play normal moves and wait for the opponent to make a mistake. He
had to provoke such mistakes by complicating games, which he did
with brio. The second factor, related to the first, is that Kasparov was
extremely well prepared for these encounters, with the exception of
the first one, as noted above. He would analyse at least 100 games
of each of his opponents to identify their strengths and weaknesses.
Then, he would use this knowledge to steer the play into the kind
of positions that a specific opponent did not handle well and thus
increase the chances for that opponent to play inferior moves. For
example, if an opponent was relatively weak in anticipating a sud-
den attack in what seems a quiet strategic position, Kasparov would
 THE BEST MOVE 41

choose an opening likely to lead to such a position, even if this meant

a slight disadvantage. In addition to creating positions where his
opponents could not play at their best, this approach also had the
advantage of reducing the search space, as some moves would not
even be considered if they did not fit Kasparov’s plan. Obviously, it
could be argued that his opponents could have used the same strata-
gem, studying Kasparov’s games to take advantage of his weaknesses.
The problem, however, is not only that Kasparov had far fewer weak-
nesses, but also that his opponents in general did not have sufficient
mastery of chess to implement such a plan.
It turns out that I played against Kasparov when he took on the
six-man Swiss national team, which was held in 1987 in Zurich. Play-
ing White, I entered a variation that he had prepared for the world
championship match against Anatoly Karpov. There was a possibility
for me to conclude the game with an early draw, by repeating the
same position three times. However, as a team we had agreed not to
accept such early draw offers, which would have made Kasparov’s
task easier, and I kept playing. With hindsight, this was a mistake,
as I rapidly lost my bearings in wild tactical complications and got
roundly beaten. In a sense, there was no game, as all the key moves
had been anticipated in Kasparov’s home preparation. The other five
games were more balanced but Switzerland got smashed 5½ – ½.
Even more impressive than the match results were the post-mor-
tem analyses. To begin with, the analysis that Kasparov made of our
strengths and (mostly) weaknesses were spot on. After analysing 100 of
our games, he had a better understanding than we had of the play of
our teammates, whom we had known for many years, and even
of our own play. In several games, he fully implemented his plan,
including the example I gave earlier of a player having a poor sense
of danger in quiet positions. In fact, in spite of theoretically having six
times less time for calculating variations than his opponents, he actu-
ally calculated deeper and better. The impression was that he was able
to generate the first moves of the key variations quickly – presumably
by using his knowledge of typical sequences of moves in similar
positions – and really started computing moves at the point where
42 T H E B E S T M O V E

we had to stop because we had reached the limit of our look-ahead

capability. In 1986, leading British grandmaster Anthony Miles, after
losing a match against Kasparov 5½ – ½, called him “a monster with
a hundred eyes who sees everything”. We came to appreciate how
appropriate this description was. Incidentally, that day I understood
that I should spend more time studying for my PhD in psychology
and less on chess!
 4
PRACTICE MAKES ALMOST PERFECT

Can anybody be a grandmaster in chess through hard practice,

or is some special talent needed? This is the classical nature vs.
nurture question, which still pervades psychology, polarising
research. The nature side claims that innate talent is needed. The
idea goes back to polymath Francis Galton, a cousin of Charles
Darwin, who among other things invented fi ngerprinting for
identifying humans. Comparing the pedigree of natural sons of
eminent men with adoptive sons of the same men, he concluded
that eminence is hereditary. By contrast, the nurture side of the
argument rejects any role for talent and contends that practice is all
that is needed. One of the major proponents of this idea was John
Watson, the father of behaviourism. In his 1930 book Behaviorism,
he made his point of view crystal clear:

Give me a dozen healthy infants, well-formed, and my own speci-

fied world to bring them up in and I’ll guarantee to take any
one at random and train him to become any type of specialist
I might select – doctor, lawyer, artist, merchant-chief and, yes,
even beggar-man and thief, regardless of his talents, penchants,
tendencies, abilities, vocations and race of his ancestors.
44 P R A C T I C E M A K E S  A L M O S T  P E R F E C T

In spite of more than a century of arguments, the debate still rages.

One reason behind the popularity of this question is that the way one
answers it has huge implications for education and training, not only
in schools but also in business and industry. Common sense would
argue that both talent and practice matter, but researchers sometimes
lack common sense! Indeed, they are still taking rather extreme posi-
tions nowadays. For example, in a famous paper emphasising the role
of deliberate practice, Anders Ericsson and colleagues argued that,
with the right kind of practice, anybody can reach the highest levels
of expertise. At the other extreme, British psychologist Hans Eysenck
argued that, unless one is talented, there is no hope of reaching high
levels in a domain. As it turns out, research on chess has much to say
on this question.

THE CASE FOR PRACTICE

Currently, the pendulum has swung to the practice side, in particular
due to the popularity of the deliberate practice framework, devel-
oped in 1993 by Anders Ericsson, Ralf Krampe and Clemens Tesch-
Römer. The central idea is that talent is not necessary to reach superior
performance, but that the right kind of practice – called deliberate
practice – is sufficient. Such practice is characterised by activities
aiming to improve performance. These activities, which tend to be
carried out solitarily, aim to correct errors and improve weaknesses
by using rapid and informative feedback. They must be carried out
for many years, and are effortful, not enjoyable. They also require
concentration and thus can be done only for a few hours a day. In
addition, they require the presence of a coach, and, in some domains
such as figure skating or swimming, access to expensive sport facili-
ties. Therefore, deliberate practice is possible only when there is a
suitable environment, including strong financial and family support.
Ericsson and colleagues emphasise both the quality and the amount
of practice and argue that, in nearly all domains of expertise, at least
10 years, or 10,000 hours, of practice are necessary for reaching top
levels. They report data from classical pianists and violinists, but make
 PRACTICE MAKES ALMOST PERFECT 45

it clear that their conclusions apply to any domain of expertise. The

role of talent is rejected, except for height, motivation and the capac-
ity to engage in deliberate practice. Importantly for our discussion
of chess skill, the authors argue that expert performance does not
depend on inherited cognitive abilities.
These ideas are certainly reasonable from chess players’ point of
view. Many players have the reputation of working extremely hard to
improve or maintain their skill level. Hungarian grandmaster Lajos
Portisch was known to analyse games eight hours a day, every day.
World champions Bobby Fischer and Magnus Carlsen said in inter-
views that they think about chess nearly all the time. A biography
of the three Polgár sisters mentioned that they practiced from 8 to
10 hours a day. These ideas are also consistent with what I have said
about the importance of chunking and knowledge in becoming a
chess expert. If players must acquire a large number of chunks to
reach high levels of skill, then it stands to reason that they will have
to devote considerable amounts of time to study and practice chess.
Another line of support for the role of practice in chess stems from
the fact that players’ overall skill level has massively increased follow-
ing four innovations in the way chess knowledge is transmitted. First,
in the 1960s, a group of Belgrade masters and grandmasters started
publishing the Chess Informant, a book originally published twice a year
(and now four times a year), which made widely available informa-
tion that was before restricted to Soviet players. The great originality
of the series was that it developed an international chess language that
basically eliminated language barriers. Second, in the 1970s, the pub-
lication of the five-volume collection of the Encyclopaedia of Chess Openings
systematised the theory of chess openings, the knowledge of which
is central for becoming a strong player. Third, starting from the late
1980s, computerised chess databases and increasingly strong chess
engines revolutionised training and tournament preparation. Earlier,
even after the advent of the Chess Informant and the Encyclopaedia of Chess
Openings, it was necessary to comb dozens of books and magazines to
find the relevant information, and then to play the games on a physi-
cal chessboard, which was time consuming. Suddenly, hundreds of
46 P R A C T I C E M A K E S  A L M O S T  P E R F E C T

games could be accessed in an eye blink, and rapidly played on the

computer screen. In addition, it became possible to use chess engines
to analyse games and, in particular, to explore and test new ideas in
the openings. Finally, the Internet made it possible to play chess 24
hours a day, seven days a week, against opponents that include many
grandmasters.
Ericsson and colleagues made their case by collecting data on vio-
linists and pianists of different levels of expertise. They also referred
to chess several times, noting that no player, including Bobby Fischer,
became grandmaster in less than 10 years of practice. Several empiri-
cal studies have tested Ericsson and colleagues’ claims. In a first study
led by Neil Charness, a colleague of Ericsson at the University of
Tallahassee in Florida, chess players were asked about the type and
duration of their practice activities. Overall, the results supported
deliberate practice. There was a strong correlation between play-
ers’ rating and the number of hours they had spent studying chess
alone. This correlation was stronger than that between rating and
the number of hours players had spent practicing or studying with
others. They also found that the presence of a coach had no effect
on skill, once the number of hours devoted to solitary practice was
controlled for statistically. Interestingly, the number of books owned
was a predictor beyond deliberate practice. Whether this result is still
valid today is an interesting question. The data were collected in 1993
and 1994, and books were then the main means by which players
acquired chess knowledge. As noted above, chess players’ practice
today consists of using databases storing millions of games, analys-
ing games with computer programmes that are vastly stronger than
world-class grandmasters and playing games on the Internet.

PRACTICE IS NOT ENOUGH

Together with my PhD student Guillermo Campitelli, I replicated
Charness’s study. Many of our findings were in line with the earlier
study, but some were at variance with the deliberate practice frame-
work. On the positive side, we found that players needed, on average,
 PRACTICE MAKES ALMOST PERFECT 47

about 11,000 hours to become a master. In addition, deliberate prac-

tice explained about 18% of the variance in skill, which is not far
from the 25% found by Charness and colleagues. On the negative
side, several results did not fit the predictions of the theory at all.
First, group practice, which included competitive games and practic-
ing with other players, had a more important role than individual
practice. Second, even though the (future) candidate masters and
masters had practiced the same amount of time during the first three
years of serious practice, the masters already had higher ratings at
this stage. This counts against the hypothesis that skill is a monotonic
function of the amount of deliberate practice. Third, the candidate
masters did not improve much after these first three years in spite
of considerable amounts of deliberate practice. Fourth, Ericsson and
colleagues argued that the most important and frequent activity train-
ing activity was to predict the next move in masters’ games, compare
one’s answer with the move actually played and receive immediate
feedback. We found that players engaged in numerous other activi-
ties, such as studying openings and endgames, finding new moves in
the opening to surprise their opponents and playing games against
humans and computers.
Finally, and most importantly, there was a huge inter-individual
variability in the amount of deliberate practice. The quickest players
needed only 3,000 hours of deliberate practice to reach master level,
while the slowest needed 24,000 hours – that is eight times longer, a
huge difference! In addition, some players devoted more than 25,000
hours to deliberate practice, but never reached master level.
This inter-individual variability, which counts against the assump-
tion that it takes 10 years to become an expert, is also reflected at the
top level. While it is true that many players needed at least 10 years of
dedicated practice to become grandmaster, others were much quicker.
Sergey Karjakin holds the record of world’s youngest grandmaster:
he obtained the title at the age of 12 years, 7 months and 0 days,
seven years after learning the rules. Current world champion Magnus
Carlsen was even quicker: he started practicing chess at the age of
8 years and became grandmaster at the age of 13 years, 3 months
48 P R A C T I C E M A K E S  A L M O S T  P E R F E C T

and 27 days – that is, only five years and four months of deliber-
ate practice! Carlsen is actually not known to be the most assiduous
worker – for example, he prefers watching or playing football to
studying chess. In a 2014 study with my PhD student Morgan Ereku,
I estimated the amount of time that the top 11 players in the world
had engaged in deliberate practice. We found that Carlsen’s number
of years of practice was significantly less than the average of the other
players. In spite of this, the difference between Carlsen and the second
player in the world (Levon Aronian) was 66 Elo points, which was
about the same as the difference between Aronian and the 14th player
in the world (Anish Giri)!

STARTING AGE, HANDEDNESS AND

SEASONALITY OF BIRTH
A major weakness of this field of research is that nearly all studies have
only used measures directly related to deliberate practice and did not
consider other possible explanations for skilled behaviour. The study
with Campitelli was an exception, as we also used several measures
unrelated to deliberate practice. The results suggested that other fac-
tors are important in reaching high levels of expertise. To begin with,
the age at which players started to practice chess seriously correlated
with skill level, even after controlling for deliberate practice statisti-
cally. The younger the players started practicing chess seriously, the
better they became. The effect was rather strong: players who started
to play seriously at the age of 12 or before had 1 chance out of 4 to
become an international-level player – not a bad deal! – but players
who started to play after the age of 12 had only 2 chances out of 100
to reach that goal. The results supported the hypothesis that there
is a critical period for acquiring chess skill, a hypothesis originally
proposed by Arpad Elo, the creator of the Elo rating.
The second factor beyond practice that we investigated was hand-
edness. Handedness is a classic marker of talent in domains such as
music, mathematics and the visual arts. Our results showed a link
between handedness and chess skill. Chess players tended to be
 PRACTICE MAKES ALMOST PERFECT 49

more often left-handed or ambidextrous than the population at large

(18% vs. 10.2%, respectively). Also, the degree of handedness was less
extreme with chess players. That is, they tended to be more mixed-
handed. However, when considering chess players only, there was no
correlation between handedness and skill level.
Finally, together with another PhD student, Philippe Chassy, I
found that there was a link between seasonality of birth and chess
skill. Chess players in the northern hemisphere tend to be born more
often in the first half of the year than in the second half. For play-
ers rated higher than 2000 points (candidate masters and above),
the respective percentages were 52.3% and 47.7%; for players at the
grandmaster level, the effect was stronger and the respective percent-
ages were 56.9% and 43.1%.
There is a seasonal pattern in several sports as well (e.g. football
players tend be born more often in the autumn). The standard expla-
nation is selective drop-out, where younger children tend to leave the
sport because they compete against peers who are a few months older
and thus are on average stronger, faster and better coordinated. How-
ever, what is exciting about the chess data is that this explanation does
not work: children playing chess compete against other children of
varying ages and even against adults. What is then the explanation? A
possibility is that the development of the foetus’s brain is affected by
external factors that tend to be present in late winter and early spring;
the likely culprits are viruses, and in particular the flu virus. Remark-
ably, the same month-of-birth pattern has been found with schizo-
phrenia, where a similar explanation has been proposed. Chapter 10
will discuss the possible link between chess and psychopathology.

TALENT IN CHESS
As noted at the beginning of this chapter, the main alternative expla-
nation to practice is talent. Theoretically, there are good reasons to
believe that there are innate differences in cognitive abilities in gen-
eral, and in those underpinning chess skill in particular. From an
evolutionary point of view, variability is necessary for evolving not
50 P R A C T I C E M A K E S  A L M O S T  P E R F E C T

only physical but also cognitive traits, as was already demonstrated

by Charles Darwin in the 19th century. In addition, going back to
chunking theory described in Chapter 2, it is reasonable to assume
that there are individual differences with respect to its parameters. For
example, one individual might need 8.5 seconds to learn a chunk, but
another only 7.5 seconds. Given the vast number of chunks that must
be learnt, even small differences could have huge effects after years
of practice. Similarly, a larger short-term memory capacity would
make it easier to learn chunks and other useful information. It would
also facilitate real-time information processing, for example when
carrying out look-ahead search. Finally, the hypothesis of individual
differences in cognitive ability is clearly supported by the substantial
literature on intelligence.
This hypothesis is also supported by child prodigies. In her PhD
thesis, Yu-Hsuan Chang reports the case of an exceptional 10-year-old
American female chess player, known only as CS, who needed little
time to become an expert. It took her only about 160 hours of indi-
vidual study and 3,800 hours of overall chess-related experience to
reach a rating of 2141. Interestingly, CS scored much better than a con-
trol group on a test of visual short-term memory. The case of Canadian
player Harmony Zhu is remarkable as well. She became world cham-
pion at age 7 in the Girls Under 8 Category. At 1545, her rating is not
that impressive; however, this is more than compensated for by the fact
that she is also an accomplished concert pianist. Before reaching the
age of 11, she had played with several renowned orchestras, such as
the Philadelphia Orchestra and the Israel Philharmonic Orchestra, and
won numerous piano competitions. Chess and music do not always go
along well together, however. “It bothers me so much”, says Harmony,
“that I can hardly concentrate on the chessboard because of my own
composed music in my head!” In spite of her talents, she remains a
normal child. Interviewed after winning the chess world title for her
age, she was asked about her coach, but could not remember whether
he was a grandmaster or even his name. Unfazed, she carried on: “Oh,
oh by the way, you know I have a pet guinea pig? . . . And his name is
Scrumptious. . . . He’s a white guinea pig”.
 PRACTICE MAKES ALMOST PERFECT 51

In recent years, the importance of talent in chess has received

strong empirical support, mostly from research into intelligence but
also from research into personality. While a fair amount of research
had been carried out for a long time on the link between intelligence
and chess, the results seemed conflicting. This was, for example, the
conclusion I reached in my book Understanding Expertise, published just
three years ago. However, two very recent meta-analyses, carried out
by a group of researchers at the University of Liverpool and Michi-
gan State University, and led by PhD students Alexander Burgoyne
and Giovanni Sala, have shed important light on this apparent incon-
sistency. (A meta-analysis is a statistical technique putting together
the results of all the studies that can be found on a particular topic.
Meta-analyses make it possible to draw more reliable conclusions than
single experiments, not only because they rely on a larger number of
participants but also because studies are replicated several times, with
different research groups and populations.)
The first study found that chess players were more intelligent, on
average, than people who do not play chess. The effect size was half
a standard deviation – a medium-size effect. The second study found
that skill in chess correlated with different measures of intelligence,
including fluid reasoning, short-term memory and processing speed.
The highest correlation was with numerical ability (0.35) while,
somewhat surprisingly, the correlation with visuo-spatial ability was
only modest (0.13). On average, the correlations were about 0.25.
Interestingly, the correlations were stronger with youth samples than
with adult samples and with unranked players than with ranked play-
ers. As we shall see in Chapter 9, teaching individuals to play chess is
unlikely to improve their intelligence, which rules out the possibility
that there is a causal link from chess to intelligence. Thus, the results
of the two meta-analyses suggest that individual differences in chess
skill are in part caused by differences in cognitive ability, in particular
with young chess players and weak players.
Given their high level of heritability, the presence of distinct per-
sonality traits in chess players would provide additional support for
the presence of innate talent. A study with my PhD student Merim
52 P R A C T I C E M A K E S  A L M O S T  P E R F E C T

Bilalić showed that children who play chess regularly tend to be

more extroverted, more open to new experiences and more prone to
arguing (as shown by their low scores on the measure of “agreeable-
ness”) than children who do not play chess. None of the measures
correlated with skill level. In addition, boys obtained lower scores on
agreeableness than girls, a finding that has been repeatedly obtained
in adult samples. Given the competitive nature of chess, this result
might suggest a reason why boys tend to be more interested in
chess than girls. In another study by Sabine Vollstädt-Klein and col-
leagues, elite chess players filled in a personality inventory. It was
found that the personality profile of male players did not differ from
the population. By contrast, female players tended to have higher
levels of satisfaction with life, to have fewer physical complaints and
to display higher achievement motivation than the population. With
males, stronger players tended to be more introvert, while, with
females, the opposite was true. Stronger female players were also
more aggressive.

INTERACTION BETWEEN PRACTICE AND

TALENT IS KEY
While research in chess psychology, like other fields of psychology,
has been highly polarised between the extreme positions of practice
and talent, it is almost certain that the truth lies somewhere in the
middle. Importantly, there will be important interactions between
practice and talent, not only at a given point in time, but also as exper-
tise develops. These ideas are captured in what could be called the
“spaghetti model” (see Figure 4.1). In this model, the environment
has an effect on practice, intelligence and performance; intelligence
has an effect on practice and performance; finally, practice and per-
formance mutually affect each other, perhaps because increasing one’s
performance boosts motivation and hence practice. A complete model
is likely to be much more complicated – for example, personality and
motivation should be included – but this model captures some of the
main features of the development of chess skill.
 Environment
Performance

Environment
Practice Intelligence

Performance
Environment

Practice Intelligence

Performance

Practice Intelligence

Figure 4.1 The spaghetti model, which emphasises interactions between environmental and innate factors.
54 P R A C T I C E M A K E S  A L M O S T  P E R F E C T

An important aspect of the model is the presence of feedback loops.

For example, intelligence affects both practice and performance, but
in turn performance affects the amount and quality of practice. A sec-
ond important aspect of the model is its dynamic nature, in the sense
that it captures the time dimension. For example, practice at time
1 will affect practice at time 2. A third important aspect is that the
model captures what is sometimes called the rich-get-richer effect,
which also relates to chaos theory in mathematics: small changes at
the beginning might amplify and have huge consequences much later.
For example, let us go back to the hypothesis that individuals might
differ with respect to the rate at which they learn a chunk. One person
might have a rate of 7.9 seconds per chunk, another 8.1 seconds per
chunk. The difference is small, but propagated over the acquisition of
100,000 chunks, the effects become considerable. In particular, the
effects are amplified by other factors. For example, a person acquiring
chunks a bit quicker will be better at chess than her friends, other
things being equal. She will win more games, which will increase
her motivation and thus the time she spends studying chess. This will
increase her strength even further.
Similar snowball effects are likely to occur with personality factors
that encourage study and training, or that give that small additional
edge in a long and difficult game. Such effects can also be caused by
acquired factors, such as better heuristics for making decisions or bet-
ter techniques for learning. Obviously, some of this knowledge might
be communicated by experienced coaches.
Finally, there is a topic that has been little studied scientifically in
chess but that is nonetheless critical: passion. Passion is a combination
of love and obsession for the game, competitiveness and rage to win.
All top-level players and most strong players have it.
 5
MEN VS. WOMEN

Women have historically been underrepresented in several fields

such as science, the arts and business. For example, there are very
few female mathematicians, and the same applies to classic com-
posers. For every Nobel Prize won by a woman in chemistry, phys-
ics and physiology/medicine, there are 35 won by men. Similarly,
males still dominate in occupations such as engineering. What are
the reasons behind this large gender gap? Several explanations have
been proposed, including biological differences, socio-cultural and
educational factors and motivational causes. In spite of the societal
importance of the topic and considerable research efforts, no agree-
ment has been reached among researchers. As we shall see, chess
sheds some important light on some of these explanations.
There is little doubt that men perform better at chess than women.
In the international rating list, there is only one woman for every
17 men. Currently, only one woman belongs to the best 100 play-
ers: Chinese grandmaster Hou Yifan, ranked at the 64th place with
a rating of 2680. Before her, the only woman to be able to com-
pete with the very best male players was Hungarian grandmaster
Judit Polgár, who was ranked number 8 in the world at the peak of
her career.
56 M E N V S . W O M E N

THE POLGÁR SISTERS

In 1965, László Polgár, a Hungarian educational psychologist, was
courting his future wife Klara, a foreign language teacher from
Ukraine. Rather than writing poetry, László used his love letters to
convince Klara of his educational views. His key argument was simple,
and actually went directly back to John Watson: there is a genius in
every healthy child, and with the correct method of education, par-
ents can grow their child into a famous mathematician, musician or
scientist. László wanted to test his theory, and thus wanted to find
a woman that would marry him and have children with him. Klara
agreed.
They had three girls: Susan, Sofia and Judit. As there was no boy,
the experiment gained additional piquancy: to show that girls are not
inferior intellectually to boys and can reach top levels of expertise.
The eldest sister, Susan, was born in 1969. László originally wanted to
run his experiment with mathematics, but switched to chess because
Susan adored the chess pieces – they were toys for her. She began
playing chess when she was 4 years old. A few months later, she
won the Budapest Girls’ Under-11 Championship, with a 100% score.
Susan became the highest-rated female chess player in the world at
the age of 15 and earned the grandmaster title seven years later.
The education of Susan, and later of her two sisters, consisted of
chess training for about 9 hours a day. In addition, she learnt math-
ematics and several foreign languages. Originally, her parents had
to fight against the Hungarian educational system, as home school-
ing was unheard of in a communist country. Chess players were not
impressed either, and were reluctant to help. However, as Susan started
getting impressive results, the family increasingly received support
from leading Hungarian chess players. László continuously refined his
training methods, and there is no doubt that Sofia and Judit (born in
1974 and 1976, respectively) benefitted from this.
By far the strongest of the three was Judit. She was the youngest
grandmaster ever when she was awarded the title in 1991, at the age
of 15 years, 4 months and 28 days old, improving on Bobby Fischer,
 MEN VS. WOMEN 57

who got the title at 15 years, 6 months and 1 day. At one point ranked
8th in the world, she has won games (in either classic or rapid chess)
against many of the top players of the world, including Anatoly Kar-
pov, Garry Kasparov and Magnus Carlsen.
Sofia did not reach as high a level as her sisters. In an interview,
Susan noted that Sofia was the most talented of the three, but also
the laziest. That being said, when Sofia was 14 years old, she won a
tournament in Rome ahead of several leading grandmasters, with the
stunning score of 8½ out of 9 and a performance of 2879 Elo, which
was at the time one of the highest in chess history.
The Polgár sisters also obtained remarkable results when playing
together for Hungary, which might be called Polgáry, as they made
up 75% of the four-player team. They won the women’s chess Olym-
piads in 1988 and 1990, breaking a long series of victories by the
Soviet Union.
László Polgár’s experiment yielded clear results: his method of
education, which emphasises specialising in a field from an early
age, produced top-level chess players. But does it really prove that
nurture is more important than nature? No, because the design of the
experiment lacks a key feature: random selection of the participants.
Thus, genetic influences are possible, even likely: László and Klara
each have a PhD degree in education. A better design would be to
raise randomly selected children. In fact, Dutch billionaire Joop van
Oosterom proposed to do just this: he suggested to László and Klara
Polgár that they could use their training methods with children from
a developing country. László was interested, but Klara declined, prob-
ably wisely. As Susan Polgar put it in a Psychology Today interview: “[My
mother] understood that life is not only about chess, and that all the
rest would fall on her lap”.

EXPLANATIONS
Gender differences in chess performance are not disputable. But what
are the reasons? Are they the same as those proposed to explain gen-
der differences in science, technology, engineering and mathematics
58 M E N V S . W O M E N

(STEM)? I briefly discuss five explanations: statistical, biological,

socio-cultural, psychoanalytical and motivational.

Statistical explanations and participation rates

A first important statistical fact is that with intelligence and mathemat-

ics tests, males display greater variability than females. That is, the
standard deviation is higher for male than female scores. Thus, extreme
values are statistically more likely with males than females. So, with
respect to intelligence, males are more likely to score either very high
or very low. A 2006 study by Chabris and Glickman shows that this
result does not hold for chess, since females’ standard deviation was
actually larger than males’. A second important statistical fact is that
there are much fewer female than male players (as seen earlier, a 1:17
ratio in the international rating list). Thus, before we discuss other
factors, we must consider the following simple statistical explanation:
when there are two groups of different sizes, the best (and the worst)
individuals are probabilistically more likely to come from the larger
group. Bilalić and colleagues showed that this explanation successfully
accounts for the gender differences in the German rating list. However,
this is only part of the explanation, as the issue seems to be postponed:
why do more boys than girls decide to play chess regularly?

Biological explanations

These are the preferred explanations by leading grandmasters. There

are two main variants, both based on partly innate traits: the first is
based on intelligence, the second on differences in aggressive behav-
iour. For example, in a 1962 interview Bobby Fischer put it bluntly:
“They’re all weak, all women. They’re stupid compared to men. They
shouldn’t play chess, you know”. In 1989, in an interview with Play-
boy magazine, Garry Kasparov was hardly more positive: “Well, in
the past, I have said that there is real chess and women’s chess. Some
people don’t like to hear this, but chess does not fit women properly.
It’s a fight, you know? A big fight. It’s not for women. Sorry”.
 MEN VS. WOMEN 59

While the phrasing was unfortunate, Fischer and Kasparov have a

point: there are gender differences with respect to intelligence and
aggressiveness, as discussed in a massive scientific literature. But the
story is more complicated than just males being more intelligent
and more aggressive. In fact, most researchers agree that there are
no gender differences with respect to overall or general intelligence.
However, there are some more subtle differences, in particular when
one considers particular cognitive abilities. Specifically, males do bet-
ter on spatial tasks; for example, they perform nearly one standard
deviation better in mental-rotation tasks. They also perform better in
mechanical tests. By contrast, females do better with respect to face
recognition, episodic memory tasks and language. For example, girls
start talking sooner than boys and already have a larger vocabulary
by the age of 3 years.
Gender differences in aggressiveness have been well-documented
in psychology. On average, men are more aggressive than women.
Men’s aggressive behaviour tends to include direct aggression leading
to physical pain and injury, while women tend to produce more indi-
rect aggression such as psychological and social damage, for example
by gossiping or spreading rumours. Men tend to use aggression as a
means to exert control over other people, while women use aggres-
sion as a means to cope with extreme stress and loss of self-control.
Gender differences occur early in childhood; for example, boys prefer
violent games, while girls are attracted by educational and entertain-
ment games.
There is much controversy about the reasons explaining gender
differences in intelligence and aggressiveness. Some researchers have
proposed that gender-specific evolutionary pressures have led to
genetic differences affecting not only brain anatomy and physiology
but also behaviour. Others have emphasised environmental factors,
such as socialisation and the kind of toys and games given to girls
and boys.
To my knowledge, the only study to have reported data on gender
differences and intelligence with chess players is the study published
in 2006 by Bilalić and colleagues. In their relatively small sample (57
60 M E N V S . W O M E N

children), boys performed better both in a chess test and an intelli-

gence test. Another study by Bilalić and colleagues, already mentioned
in Chapter 4, supported the hypothesis that boys are more aggressive
than girls: in their sample, boys scored lower on agreeableness than
girls. In addition, chess players scored lower than children who did
not play chess.

Socio-cultural explanations

A standard explanation for gender differences in STEM disciplines

and business is sexism and what is sometimes called the glass ceil-
ing: an invisible but unbreakable barrier that makes it impossible
for women and other minorities to progress to the highest levels of
their field. Whether this explanation applies to chess is an interesting
question. On the one hand, chess competitions are fair and open to
anybody, assuming one has the required skill level. On the other hand,
discrimination may be subtler. For example, the kind of comments
made by Fischer and Kasparov certainly do not look like a welcoming
message to women, nor does the fact that there are typically very few
women in chess clubs.
Another socio-cultural explanation is that women, to some extent,
are victims of an inferiority complex. They believe than men are bet-
ter, perhaps because they think that chess corresponds to the type
of intelligence (e.g. spatial and mathematical intelligence) that soci-
ety assigns to males. When they face men, they may have smaller
expectations and assume that they are weaker, and tend to confirm
this prediction by losing more often than should be the case based
on chess skill only. A variant of this hypothesis is stereotype threat,
where the awareness of a negative stereotype negatively affects one’s
performance.
This hypothesis has received mixed support. In an experiment on
stereotype threat carried out by Anne Maass and colleagues, partici-
pants played against an opponent via a computer. There were three
conditions: (a) control group, with no knowledge of the opponent’s
gender; (b) high gender stereotype, where participants were told
 MEN VS. WOMEN 61

the opponent was a man; and (c) low gender stereotype, where par-
ticipants were told that the opponent was a woman. In the last two
conditions, the gender stereotype was made explicit by telling players
that men play better chess than women. Women performed worse than
men in the high stereotype condition, but not in the other two con-
ditions. However, the hypothesis of stereotype threat was not sup-
ported in another study. In an analysis of over 5.5 million competitive
games in international tournaments, Tom Stafford did not find any
support for the stereotype threat hypothesis. Contrary to expectation,
women actually performed better against men than expected from
their ratings.

Psychoanalytical explanations

This is perhaps the less likely explanation, but it is worth discussing

in some detail, as it has been widely publicised. For example, it was
discussed in detail by grandmaster Reuben Fine, who, before becom-
ing a Freudian psychoanalyst, was one of the contenders to the world
championship title. (When Fine decided to leave chess to become a
psychoanalyst, a fellow grandmaster joked: “a great loss to chess, and
at best a draw for psychoanalysis”.)
The theory, originally proposed by Ernest Jones, Sigmund Freud’s
official biographer, aimed to explain two sets of data: first, there are
few female chess players, and second, several top-level players suf-
fered from psychiatric disorders (see Chapter 10). According to the
theory, women play badly because they do not want to play well!
Chess is a substitute for war, and is characterised by anal-sadistic fea-
tures. The familial constellation is symbolically reflected in chess: the
father is the king, the mother is the queen and the child is the pawn.
Thus, chess embodies the Oedipus complex. The male child wants to
kill the father (“checkmate” means “the king is dead” in Persian) to
seduce the mother. However, if the child is a girl, she wants to get rid
of the mother to seduce the father. Therefore, there is no interest for
a girl to checkmate the king, hence to kill him, since it is precisely
he who is the subject of her seduction. As a consequence, women do
62 M E N V S . W O M E N

not excel in chess. According to Freudian logic, the women who do

excel in chess have a reverse Oedipus, where the girl sides with her
mother to kill the father.
The theory has received little empirical support. For example, no
systematic survey has tried to test Jones and Fine’s hypotheses. Kro-
gius, in his book the Psychology in Chess, notes that Freudian assumptions
are “hard to believe” and regrets that Fine did not use his experience
as a top-level grandmaster. I certainly agree with this evaluation. One
could also mention that the idea of a reverse Oedipus is actually a
clear example of an ad-hoc explanation to save a theory.

Motivational explanation: women have broader interests

A final explanation is that chess-playing women have broader interests

than men. For example, they yearn for a recognised job, social and
family life, and even spiritual development. Therefore, they are less
likely to be obsessed by a single activity such as chess. This explana-
tion is supported not only by interviews given by top female players,
but also by their actual career choices.
Examples abound. Former world champion Maia Chiburdanidze
studied medicine and engaged herself in social work. Grandmaster
Dana Reizniece-Ozola, who represented Latvia several times in the
Chess Olympiads, gave up chess to become Minister of Finance of her
country. Hou Yifan, who became the women’s world champion at the
age of 14 and is currently the only woman to successfully compete
with the best male players, studied international relations at Peking
University, to the great despair of her coach. According to her, “chess
is just a game; chess is not life”. In 2018, she won a prestigious Rhodes
Scholarship to study for an MSc in Education at the University of
Oxford. To be clear, the highly competitive scholarship was awarded
based on her academic performance and not her chess career. Previous
well-known Rhodes Scholars include former US president Bill Clinton
and former US National Security Advisor Susan Rice.
Some male players have, of course, also had careers beyond chess –
good examples are Reuben Fine, whom I have discussed above, and
 MEN VS. WOMEN 63

Kenneth Rogoff, an American grandmaster who became professor

of economics at Harvard University and was chief economist of the
International Monetary Fund. And of course, former world champion
Garry Kasparov, who retired from chess for a (mostly unsuccessful)
career in politics in Russia. However, the point is that women propor-
tionally move more often to other careers than men.

REDUCING THE GENDER GAP

Although the presence of gender differences is undeniable in chess,
the reasons behind them are poorly understood. Just like with STEM
disciplines, biological and socio-cultural mechanisms are likely to
be involved and to interact in complex ways. The explanation based
on a glass ceiling effect is not as convincing as in other fields, partly
because promotion in chess is based on highly objective measures –
essentially, winning, drawing or losing a game. This being said, subtle
factors might discourage girls and women to engage in chess. The
least likely explanation seems to be that based on psychoanalysis.
There have been many initiatives, both personal and institutional,
to improve women’s skill level in chess. For example, some tour-
nament organisers have tried to improve women’s motivation by
offering equal prizes for the two genders. There has been a trend to
encourage mixed tournaments at the international level, by inviting
more female players. At the local level, chess clubs have tried to create
a more attractive atmosphere for women by getting rid of stereotypes.
At the personal level, top female players such as Judit Polgár and
Hou Yifan have systematically avoided female-only tournaments –
sometimes even spurning the female world championship – in order
to challenge leading male grandmasters and gain valuable experience.
 6
STYLE AND INTUITION

In chess, style concerns the characteristics of one’s knowledge and

thinking that affect the kind of moves one chooses. One of the attrac-
tions of chess is that it is possible to play with widely different styles.
There is no standard way of describing a player’s style, but the follow-
ing dimensions are often used: tactical (combinational) – strategic
(positional); aggressive – defensive; risky – safe; and classic – original.
Other classifications characterise styles as technical, dynamic and
tricky. Style correlates with the number of moves that are normally
anticipated: a tactical style requires deep calculations, whilst a strate-
gic style calls for shallower calculations. Whatever their style, players
need competency in most aspects of chess. For example, a player with
a technical style but little tactical knowledge would be easy prey for
any opponent that has identified his or her weakness in dealing with
combinations.

THREE PROTOTYPICAL STYLES

Chess books tend to focus on three styles: psychological (where one
plays the opponent rather than the position), positional and tactical.
This section provides a description of these three styles with examples
of top players who used them.
66 S T Y L E A N D I N T U I T I O N

Emmanuel Lasker, world champion from 1894 to 1921, was one of

the earliest and best proponents of the psychological style. He argued
that, in order to win in chess, one had to understand one’s opponents,
his strengths and weaknesses, his likes and dislikes. The game is a fight
between two personalities: “It is two human beings who fight on the
chessboard, not the wooden pieces”. Thus, one should play moves
that are the most unpleasant for a specific opponent, rather than the
objectively best moves. Garry Kasparov is a later world champion that
excelled in the psychological style, as noted in Chapter 3, although he
could also be categorised as a combinational player.
The positional style consists of playing sound moves that avoid cre-
ating weaknesses in one’s position. Rather than starting a direct attack,
which often leads to strategic concessions, positional players prefer
to defend temporarily. One of the greatest proponents of this style
was the Cuban José Raúl Capablanca, who was world champion from
1921 to 1927. Capablanca excelled in converting small positional or
material advantages, playing with extreme precision – he was called
the “chess machine” at the time. His style was highly economical,
eschewing any superfluous complications. In spite of playing moves
that looked harmless, Capablanca was a master in coordinating his
pieces, exploiting their potentialities to their maximum. While not a
tactical player per se, Capablanca was also able to calculate very pre-
cisely when necessary. World champions Anatoly Karpov and Magnus
Carlsen are other highly skilled representatives of this style.
The combinational style focuses on direct attacks, drawing on tac-
tical ideas to implement plans. One of the best examples of this style
was Russo-French Alexander Alekhine, who held the world crown
from 1927 to 1935 and from 1937 to 1946. Alekhine’s play was
very dynamic, trying to build direct attacks against the opponent.
Of course, he also took sound positional principles into account,
in particular making sure that his initiative was worth the strategic
weaknesses that he conceded. Alekhine’s genius was to display great
imagination and risky play whilst at the same time remaining objec-
tive. Other world champions that adopted the combinational style are
Mikhail Tal and Garry Kasparov.
 STYLE AND INTUITION 67

In chess history, many players developed styles that were truly

universal. If the position was calm and called for manoeuvring, they
would play smooth positional play. If the position was dynamic and
tactical, they would calculate deeply and accurately. If they were fac-
ing an endgame, they would display flawless technique. If the posi-
tion called for psychological warfare, they would play the opponent.
World champion Boris Spassky and grandmaster Fabiano Caruana,
currently number 2 in the world and the youngest grandmaster from
both the US and Italy, are examples of this universal style. With the
advent of chess engines and chess databases, which have raised the
overall level of play, increasingly more players have developed such
a style, as being limited in some aspects of the game is just too risky
from a competitive point of view.

DOES ONTOGENY RECAPITULATE PHYLOGENY?

Swiss psychologist Jean Piaget famously proposed in his theory of
genetic epistemology that, with respect to knowledge, ontogeny par-
allels phylogeny. That is, during development children go through the
same cognitive structures as their human ancestors during evolution.
Interestingly, world champion Max Euwe proposed the same idea
with chess. In his book The Development of Chess Style, he argued that the
development of a chess player reflects the historical development of
chess styles. Thus, players go through several stages, which can be
mapped onto leading players of the time.
After an initial period where players moved their queen and other
pieces rather erratically, hoping for a decisive opportunity (Gioachino
Greco, 1600–1634), players realised the importance of pawns, their
strengths and weaknesses (François André Danican Philidor, 1726–
1795). Two combinational periods followed. First, combinations for
their own sake, where beauty and attacking ideas dominated over
material (Adolf Anderssen, 1818–1879). Then, combinations played
for crowning deep positional play (Paul Morphy, 1837–1884). The
next stage focused on strategy, and is the period when a sophisti-
cated theory of positional play, now known as classical style, was
68 S T Y L E A N D I N T U I T I O N

almost single-handedly elaborated by Wilhelm Steinitz (1836–1900).

Key principles include the occupation of the centre, the avoidance of
pawn weaknesses and the accumulation of small advantages.
This was followed by a period where positional play got mastered
to the point where it became pure technique (1900–1914; José Raúl
Capablanca is the best example of this period). In reaction to this stage
dominated by technique but also dogmatism and a lack of creativity,
a number of grandmasters came up with revolutionary ideas, which
essentially violated the key principles laid out by Steinitz. Key actors
in this period (1919–1940) include Richard Réti, Alexander Alekh-
ine and Aron Nimzowitsch. Euwe’s final stage is called the Russian
School, and starts in 1945. It is characterised by an emphasis on the
initiative and dynamic piece play, fighting spirit, counter-play and
active defence.
Euwe’s thesis is intriguing but unfortunately not developed in
great depth. It seems reasonable for the early stages of a player’s career.
Beginners do enjoy wandering with their queen, giving checks and
grabbing material. And this is rapidly followed by a stage where they
realise – thanks to their teacher, books, videos or painful defeats – that
pawns matter and that their structure dictates much of the game. At
this point, I believe that Euwe’s thesis falls apart: styles diverge from
rather than parallel chess history. Some players focus on dynamic and
combinational play; others on positional and technical play; and still
others try to have a more universal style. Nowadays, style is mostly
determined by the kind of openings played, and it is not uncom-
mon for a player to have a defensive positional style with Black and
an aggressive combinational style with White. In addition, opening
theory has developed to such a point that moves are determined
more by knowledge than by style. Even within the same opening,
and depending on the variation chosen by the opponent, a player
will alternatively play positionally, tactically and technically. Finally,
the good old times where mastering one style was enough to reach
the international level are over: the overall level of chess has increased
to such an extent that at least good knowledge and know-how of the
main styles is essential for competing professionally.
 STYLE AND INTUITION 69

A different approach to style might be mentioned here. Using

psychoanalysis, Reuben Fine analysed the styles of world champions
with respect to their personality. He concluded that the relationship
is not straightforward. In some cases, players with an aggressive per-
sonality do attack on the chessboard (e.g. Alekhine). In other cases,
it is the opposite: players who are aggressive in life defend in chess
(e.g. Steinitz).

A COMPUTER STUDY OF STYLE IN CHESS

In a controversial paper written in 2006, Guid and Bratko used com-
puter analyses to find out who was the best player ever. They used
a slightly modified version of Crafty, a very strong computer pro-
gramme, to assess the quality of the moves played in world cham-
pionship matches, starting from the 1886 match between Zukertort
and Steinitz. Fourteen world champions, from Steinitz to Kasparov,
disputed these matches. Discarding the first 11 moves of each game,
as they tend to rely on chess theory – which has improved over the
years – Guid and Bratko calculated the difference in quality between
the move proposed by the computer programme and the move actu-
ally played in the game.
The results showed that the players choosing the moves the closest
to Crafty’s were Capablanca, Kramnik and Karpov. These three players
were also, with Petrosian, the players with the lowest blunder rate.
By contrast, the three players who were the farthest from Crafty’s
choices – Botvinnik, Euwe and Steinitz – were also the three play-
ers with the highest blunder rate. With respect to style, Steinitz, Tal
and Fischer had the highest position complexity on average, while
Petrosian, Spassky and in particular Capablanca managed to keep their
games at a fairly low level of complexity. Guid and Bratko concluded
that the latter players, who can be described as calm and positional,
tended to play simpler positions, and thus on average played better
moves and avoided blunders. When a measure combining complexity
and error was used, the most precise players turned out to be Kram-
nik, Capablanca and Karpov, in that order.
70 S T Y L E A N D I N T U I T I O N

Whilst thought-provoking, this study has raised a number of criti-

cisms from the chess community. One of them is that Crafty was not
the best computer programme at the time, and indeed was weaker,
based on the Elo rating, than some of the world champions investi-
gated. Another is that, even with current computer programmes that
are vastly superior both to the version of Crafty used in that study and
the best human players, there is sometimes a fair amount of disagree-
ment in the way they evaluate a move, in particular when the position
is complex. This brings us to the third point: it is clear that Crafty did
not “understand” some of the most brilliant human moves, which
relied either on calculations deeper than what Crafty could perform
or on subtle positional considerations. There also seems to be a bias
in favour of positional players, who prefer calm positions in which it
is difficult to commit a serious error. Fourth, the style of some players
changed during their career; a notable example is Steinitz, who played
wild tactical games in his youth before adopting a defensive style
in the second part of his career. Finally, this analysis totally ignores
psychological aspects. Given their style, players such as Lasker, Tal
and Botvinnik were often playing more the opponent than the posi-
tion itself, with obvious success but also with many errors from an
objective point of view. In that respect, Siegbert Tarrasch might have
had the last word: “One doesn’t have to play well, it’s enough to play
better than your opponent”.

WHAT IS INTUITION?
Intuition can be defined as the ability to understand a situation rapidly
and effortlessly. It is a topic that has been of great interest both to
psychologists and chess players. The former consider it as one of the
defining features of expertise – think of a medical doctor able to diag-
nose a disease almost immediately. The latter wonder how a player
like Magnus Carlsen is so good at finding good moves, often with-
out much calculation, while weaker grandmasters cannot find them
despite considerable thinking. Also, how is it that the same concept –
intuition – is used to describe players with styles as opposite as world
 STYLE AND INTUITION 71

champion Tigran Petrosian (who excelled in calm, solid and strategic

positions) and world champion Mikhail Tal (who relished wild com-
binational play crowned by risky sacrifices)?
According to psychologists, experts’ intuition is characterised by
several features: speed and ease of understanding the problem and
finding a solution; perceptual and holistic nature of that understand-
ing (i.e. the problem is understood as a whole); absence of awareness
of how the solution is reached; and presence of emotions. A final
requirement, to avoid the case where people just guess, is that intu-
itions must be correct most of the time. In most domains, experts dis-
play intuitions with routine problems. However, as problems get more
complex and harder, experts tend to revert to problem-solving methods
that are non-intuitive and rely on conscious problem-solving.
Chess has been the main topic of study behind many ground-
breaking works on intuition. In his PhD thesis, de Groot emphasised
the importance of intuition, and actually argued that it is part of
most of our thinking, even with difficult problems, which at first
sight seem to be addressed with conscious, analytical methods. In the
1960s, Oleg Tikhomirov stressed the link between intuition and emo-
tions. It is perhaps surprising that this connection was first made by
Tikhomirov, a psychologist who was Soviet, Marxist and materialist.
Finally, as we have seen in chapters 1 and 2, Simon and Chase argue
that intuition is pattern recognition, and provided a theory of how
pattern recognition was made possible by the acquisition of a large
number of chunks.

FIVESTAGE THEORY OF INTUITION

Chess was also used extensively in Mind over Matter: The Power of Human
Intuition and Expertise in the Era of the Computer, an influential book written
by philosopher Hubert Dreyfus and mathematician Stuart Dreyfus.
The Dreyfus brothers proposed that the road to expertise consists
of five stages. In the novice stage, information about domain-related
facts, features and actions is acquired through instruction. The appli-
cation of this knowledge is fairly mechanical, and particularities of
72 S T Y L E A N D I N T U I T I O N

the environment are ignored. After considerable concrete experience,

individuals move to the next stage, the advanced beginner stage. The context
becomes increasingly significant and used for making decisions. In
the competence stage, efficiency increases and decision-making methods
are coordinated hierarchically. However, planning is essentially still
conscious and intentional. Intuitive understanding makes its appear-
ance in the proficiency stage, and individuals learn to pay attention to
salient features and ignore others. However, decisions still depend on
analytical thinking. It is only in the final expertise stage that decisions are
intuitive. At this stage, according to Dreyfus and Dreyfus, experts do
not make decisions, they just carry out the right action. Except for
very difficult problems, behaviour is entirely fluid. Think of Neo in
The Matrix knocking down an army of Smiths as if he was walking in
the park, and you get the picture.
While the theory does describe important characteristics of the
development of expertise (e.g. importance of context, progress from
hesitant and error-prone to fluid behaviour and of course the role of
intuition), it is seriously flawed, as I have shown in an article writ-
ten with my PhD student Philippe Chassy. An important part of our
analysis relies on chess. Dreyfus and Dreyfus argue that, at the expertise
level, conscious problem-solving does not play any role. Even when
grandmasters deliberate, for example when they face a difficult prob-
lem, they do not anticipate moves and carry out other types of analyti-
cal thinking, but rather they reflect on their intuitions. This is a very
surprising statement. How about the substantial amount of research,
starting with de Groot’s work in the 1940s, supporting the presence of
look-ahead search? A few years ago, I attended a workshop on exper-
tise held in Hubert Dreyfus’s honour, and asked him the question. He
answered that he was not aware of this research. Rather disappointedly,
one of the most cited works on expertise and intuition got it totally
wrong with respect to the empirical evidence on chess thinking!
Dreyfus and Dreyfus did mention an interesting informal experi-
ment. International master Julio Kaplan played rapid games against a
weaker master whilst adding dictated numbers at the same time. In
spite of the interfering task, Kaplan “more than held his own”.This was
 STYLE AND INTUITION 73

taken as evidence that experts can still display fluid behaviour despite
lacking time for planning. This is a reasonable conclusion, although it
hard to evaluate the experiment given the few details provided. Unfor-
tunately for the theory, when a proper experiment was carried out by
Robbins and colleagues, where players had to generate random num-
bers while solving tactical chess problems, it was found that players
were affected by the interfering task. With the group of stronger play-
ers, performance dropped by about one third when compared to the
control condition where they simply solved chess problems.

COMPUTERS AND INTUITION

In a previous book written in 1972, What Computers Can’t Do, Hubert
Dreyfus had provided a blistering criticism of the then nascent field of
artificial intelligence. His key argument was that, while computers can
reach decent levels by analytical means (such as anticipating a large
number of moves in chess), they would never be able to reach genu-
ine expertise, because they lack intuition. Developments in artificial
intelligence have proven Dreyfus wrong. Checkers world champion
Marion Tinsley was beaten by Chinook in 1994. A similar fate has
met world champions in other board games: in chess, Garry Kasparov
was beaten by Deep Blue in 1997; in Othello, Takeshi Murakami was
beaten by Logistello in 1997; finally, in Go, which was considered out
of reach of artificial intelligence, Lee Sedol was beaten by AlphaGo
in 2016. In all these domains, computers are now vastly superior to
humans. For example, in chess, the current number one player in the
world Magnus Carlsen has an Elo rating of 2843 whilst the best com-
puter programme, Stockfish 9, has a rating of 3450. With a difference
of about 600 points, or 3 standard deviations, Carlsen is expected to
score only 3 points out of 100 games.

Computer intuition

Progress in artificial intelligence has two important implications for

our understanding of human intuition. First, it raises the possibility
74 S T Y L E A N D I N T U I T I O N

that computers not only play chess better than humans, but that they
understand it more deeply and have a much better intuition. This
is a repulsive thought for most human players, but it is probably a
reality now. Most computer programmes playing chess are based on
powerful evaluation functions, combining dozens of features, such as
safety of the king, mobility of the White-square bishop and control
of centre, with the ability of searching billions of positions before
selecting a move. Still, it is totally out of the question to calculate all
possibilities, and thus computers rely on some form of “intuition” to
make a move. In fact, there are many examples of surprising moves
played by computers, which turned out after extensive human analy-
ses to be very deep positionally. In 2006, world champion Vladimir
Kramnik lost 4–2 against Deep Fritz, which actually was not even the
best computer programme at the time. In the 6th game, Deep Fritz
played a rook manoeuvre that pundits derided as child-like. The next
moves of the game showed that this manoeuvre was the prelude
to a very deep plan that led to fatal weaknesses in Kramnik’s king’s
side and ultimately to material loss.
If there were any doubts that artificial intuition is possible, these
were recently shattered by AlphaGo’s victories in Go. Just before the
entrance of AlphaGo into the Go scene in 2015, artificial intelligence had
progressed very slowly with this game, to the point that some
experts thought that computers would never be able to beat the best
humans. The most optimistic researchers were of the view that it
would take a computer at least 10 years to win a game against a
Go professional player. Thus, before the match against South Korean
grandmaster Lee Sedol, spirits were high in the Go community and it
was expected that AlphaGo would be beaten easily. After all, Lee Sedol
was only one of two players to have won 18 international titles (more
or less the equivalent to Grand Slam titles in tennis). Unfortunately
for humankind, Lee was thrashed 4–1.
AlphaGo primarily used three artificial intelligence techniques:
deep learning, reinforcement learning and Monte Carlo tree search.
Deep learning consists of sophisticated methods for adjusting the
weights of an artificial neural network, using grandmaster games as
 STYLE AND INTUITION 75

input. Reinforcement learning further adjusts these weights, taking

into account the result of a game following a given move. Monte
Carlo tree search, in a nutshell, generates games by randomly picking
moves for the two players. This is quite different from the technique
traditionally used in computer chess (e.g. Deep Blue or Deep Fritz),
where possible moves are explored in a systematic way. The rationale
is that, when averaging a large number of games, better moves in the
current position lead to better results even with random games.
AlphaGo was undoubtedly a milestone in artificial intelligence
research, but that was only the beginning. An improved version beat
Chinese grandmaster Ke Jie, the world’s No. 1, with a 3–0 score,
and won all sixty online games it played against world-class players.
About AlphaGo’s strength, Ke Jie said the following: “After human-
ity spent thousands of years improving our tactics, computers tell us
that humans are completely wrong. . . . I would go as far as to say not
a single human has touched the edge of the truth of Go”. As if this
was not enough, a new version of the programme, called AlphaZero,
learns from scratch and improves by playing against itself. That is,
except for the rules of the game, it creates all its knowledge. Alpha-
Zero is superior to the previous versions; for example, it won 100–0
against the version that beat Lee Sedol. The rating of this version is a
stunning 5000 Elo, compared to the 3700 Elo obtained by the best
human Go player. (Note that ratings in Go are not directly comparable
to ratings in chess.)
In an unpublished paper, the DeepMind team claimed that a gen-
eralised version of AlphaZero beat Stockfish, one of the best com-
puter programmes in chess, and Elmo, the leading programme of
shogi (Japanese chess), needing only 24 hours of learning to achieve
a “superhuman level of play”. The chess community has reacted
strongly to this news, arguing that the match against Stockfish was
unfair. For example, AlphaZero ran on dedicated hardware, and thus
had much more processing power than Stockfish and Elmo, which
ran on PCs. In addition, each programme had one minute to make
a move, whilst Stockfish has been designed to optimise its time use,
playing rapidly for obvious moves and thus saving time for more
76 S T Y L E A N D I N T U I T I O N

complex moves. These criticisms might be true, but the fact is that we
have now a super-program that learns by itself from scratch and that
plays much better than the best humans. In addition, AlphaZero has a
very intriguing and intuitive style, sacrificing material for long-term
positional compensations that most humans cannot see. Just like Go,
AlphaZero seems to show that human understanding of chess, in spite
of centuries of practice and study, is rather limited.

Lessons for human intuition

Progress in computer science and artificial intelligence gives scientists

powerful tools for studying intuition and indeed rationality. An ear-
lier try in this direction was carried out by Peter Jansen in 1992. He
took advantage of the presence of chess endgame databases that have
perfect knowledge of the value of each move and thus know what
is the outcome of a position with perfect play on both sides. (At the
time of writing, such databases have been built for all endgames with
up to seven pieces.) Jansen found that even the best humans played
poorly – as compared to optimal play – in simple endgames. For
example, winning the endgame king-queen versus king-rook is con-
sidered trivial and is hardly discussed in textbooks. Jansen’s results
showed that world-class grandmasters committed so many mistakes
that they needed, on average, four times longer than necessary to win
the game. Because of the 50-move rule – which states that a game is a
draw if 50 consecutive moves for White and Black are played without
the capture of a piece or the movement of a pawn – they would have
in many cases failed to win the game against the best defence.
 7
ERRARE HUMANUM EST

To err is human. This certainly applies to chess players. Errors, includ-

ing horrible howlers, are fairly common in chess. Most players,
including world champions, blunder every so often, and the techni-
cal literature on chess provides some spectacular examples. In the
game Petrosian against Bronstein in 1956, played in the Amsterdam
qualification tournament for the world championship, future world
champion Petrosian was winning but did not see that his queen was
attacked. When Bronstein took her, Petrosian resigned on the spot. In
the game Alekhine against Euwe, played in the world championship
return match in 1937, both players overlooked a simple combination
for three moves in a row. More recently, during the 2017 Wijk aan Zee
tournament, world champion Magnus Carlsen missed a checkmate in
three moves in his game against Anish Giri and let his lucky opponent
escape with a draw. Worst of all, in his match against computer pro-
gramme Deep Fritz, world champion Vladimir Kramnik overlooked
a mate-in-one and was painfully checkmated.
Relatively little scientific research has been carried out on errors
in chess. De Groot’s thesis has some mentions in passing and a few
studies have used blunders, typically identified by chess computers,
as a dependent variable. The only detailed qualitative analysis trying
to understand the mechanisms leading to errors was carried out by
78 E R R A R E H U M A N U M E S T

Pertti Saariluoma. He induced errors by choosing combinations that

could be found only by identifying two themes in close succession.
As expected, most of the errors were made with the move that linked
the two themes. Similar results were obtained with strategic posi-
tions and endgames. Saariluoma concluded by arguing that working
memory overload explains errors only in part, and that it is neces-
sary to develop more sophisticated explanations based on pattern
recognition, planning and problem restructuring. This study explored
an unchartered topic and called for more experiments; for example,
one could try to induce specific errors as a way of testing different
theories of chess skill.
By contrast, there are numerous discussions of errors in chess,
and how to avoid them, in the practical chess literature. Alexan-
der Kotov’s Think like a Grandmaster and Nikolai Krogius’s Psychology
in Chess, which both present instructive and thought-provoking
examples from grandmaster play, are good examples of such books.
While their discussions are based on observations, anecdotes and
a fair amount of speculation – explanations are often based on
introspection, hindsight and a posteriori accounts – the described
errors are real.
Kotov and Krogius provide several explanations for errors in chess.
Some of the explanations are specific to chess, whilst others apply to
other endeavours as well. Let us start with the domain specific ones.

AUTOMATISMS
Automatisms are important in chess, since they allow one to make
decisions rapidly. Some consist of very simple conditioned actions,
such as “if your opponent takes one of your pieces, take it back imme-
diately”, or “if a piece is attacked, move it away”. Others are subtler
and directly link to the chunking mechanisms discussed in chapters 2
and 3. For example, given a certain pawn structure, it is often a good
idea to place one’s knight in front of an isolated pawn. The beauty
of chess is that it is a game of exceptions, and even the best proce-
dural knowledge will be incorrect every so often. If one follows one’s
 ERRARE HUMANUM EST 79

intuition without doubling-checking variations concretely, there is

always a risk that some tactical blow will be overlooked.

CHESS IMAGES
Krogius spends a great deal of time with what he calls “chess images”.
The term is maybe not the best one, as these images contain not only
visual information on the location of pieces, but also more concep-
tual information such as the evaluation of the position. In some of
Krogius’s examples, images really refer to what I have called sche-
mas or templates in previous chapters. Chess images are very use-
ful in most cases, as they provide much information. They can also
produce powerful negative effects, however. For example, facing an
unexpected move by the opponent, a player will often still think
using the mind-set provided by a chess image that was suitable just
one move ago, and might miss a new opportunity. Several examples
will be provided below.
Another class of errors, which Krogius calls “retained images”,
concerns the case where, when anticipating a sequence of moves, a
piece taken by the opponent somehow still remains present in the
mind’s eye. The difficulty thus resides in correctly updating the loca-
tion of pieces or their disappearance in the mind’s eye. For example,
with this type of error, a player would defend in his calculation against
a bishop that actually is not on the board anymore. In a related type
of errors, which Krogius calls “inert images”, the player automati-
cally carries a positive evaluation reached at some point of the game
on to the following moves and is essentially on an automatic pilot
mode. As a consequence, the difficulty of winning the game can be
underestimated and one’s sense of danger blunted. So, for example
in the game Petrosian against Korchnoi in the 1963 Moscow tourna-
ment, Petrosian had had a winning position since the opening. Still
winning after 34 moves, he overlooked a simple combination and
played a careless move, which lost immediately. This type of error is
related to overconfidence, which I shall discuss towards the end of
this chapter. It is also related to the Einstellung effect that I described
80 E R R A R E H U M A N U M E S T

in Chapter 3. In this respect, Krogius reports a very interesting com-

ment by Petrosian: “Personally, I am convinced that if a strong master
does not see such a threat at once he will not notice it, even if he
analyses the position for twenty to thirty minutes”.
Many errors are due to attention lapses. A typical example is
the case where one player misses an obvious threat, such as in the
example above where Petrosian lost his queen in one move. To some
extent, it could be argued that at least some of the errors that Kro-
gius imputed to chess images are in fact attentional errors. In some
cases, it seems that the effort required for anticipating moves wipes
out from working memory important information, such as an attack
against one’s queen.

EMOTIONAL FACTORS
It might come as a surprise for non-players, but chess can be a very
emotional game. To begin with, most players have their bêtes noires,
opponents against whom their score is much less than expected by
their relative objective strength. For example, Soviet grandmaster
Efim Geller was the bête noire of Bobby Fischer, with five wins, three
defeats and two draws, in spite of Fischer’s superiority. Similarly,
American grandmaster Hikaru Nakamura, currently number 8 in the
world, struggles against Magnus Carlsen, having lost 12 games and
won only one, with 21 draws. Krogius reports that he analysed 80
games between 10 pairs of players where one player was a bête noire
for the other. He found that the players on the losing side commit-
ted more obvious strategic errors and tactical mistakes against their
bête noire than against other players. Krogius argues that a negative
emotional state strongly diminishes these players’ vigilance and ner-
vous resistance, although it could be argued that, in addition to this,
there might be a profound incompatibility in style. For example, in
the case of Fischer and Geller, Geller was known for his very danger-
ous, uncompromising attacking style. Interestingly, bêtes noires are
not transitive. For example, ignoring draws, Tal had a lifetime score
of 8–5 against Bronstein, and Bronstein had a lifetime score of 7–4
 ERRARE HUMANUM EST 81

against Keres. However, rather than Tal dominating Keres, as predicted

by a transitive relation, it is Keres who led by 8–4!
The opponent’s behaviour is another important emotional factor
that might cause distraction and irritation. The noise he is making
when drinking his coffee, the way he is looking at you, his after-
shave . . . pretty much anything might exasperate you. In the 1959
candidates’ tournament, Hungarian grandmaster Pal Benko accused
future world champion Mikhail Tal of hypnotising him, and decided
to wear dark eyeglasses when playing against him. (This is another
example of a bête noire, as Benko had lost the last five previous games
against Tal.) The game ended in a draw, which is all what Tal needed
to qualify for the world championship match against Botvinnik. After
the game, Tal quipped: “When I want to win against Benko I win;
when I want to draw, I draw!”
Emotional hatred and accusations of hypnosis were taken to new
heights in the 1978 world championship match in Baguio between
Viktor Korchnoi and Anatoly Karpov. Karpov included in his team
Vladimir Zukhar, who at the time was a renowned parapsycholo-
gist and hypnotist in Soviet Union. Korchnoi complained and started
wearing reflecting glasses. (This is only one of the many bizarre
incidents that single out this world championship.) Whether or not
hypnosis really played a role, I doubt it. But it is true that Korchnoi –
although admittedly playing some brilliant chess at times – made
terrible mistakes in some games. For example, in the 17th game, he
committed a string of errors which turned a winning position into
a drawn position, in which he blundered and overlooked a simple
checkmate in three moves.
A final source of emotional pressure is one’s position in the tour-
nament and the possibility of achieving an important result. In 1988,
I was playing the tournament of my life in Biel and was very close
to obtaining a grandmaster norm, a key requirement for becom-
ing a grandmaster. I would have been the first Swiss player to do
so. My opponent was former women’s world champion Nona Gap-
rindashvili. After some energetic and precise play in the opening
and early middle game, I obtained a winning position against her
82 E R R A R E H U M A N U M E S T

unconventional defensive system. However, at the crucial moment, I

missed a simple move that meant winning straight away, progressively
lost the thread and ultimately had to resign. There is no doubt in my
mind that my defeat was due to the fact that I could not handle the
pressure of being so close to a grandmaster norm. How to regulate
emotions is of course an important topic in other sports as well, and
is indeed a standard topic in sports psychology.

INSUFFICIENT KNOWLEDGE
Many errors are caused by a lack of knowledge. At weaker levels, this
can be due to not knowing typical openings, standard tactical and
strategic patterns in the middle game and common manoeuvres in
endgames. As noted in an earlier chapter, masters and grandmasters
often win games by waiting for mistakes, and lack of knowledge is
at the core of many of them. Mistakes due to insufficient knowledge
also occasionally occur at the top level. For example, in the first game
of the world championship held in 2010 in Sofia between Veselin
Topalov and Viswanathan Anand, world champion Anand could not
remember the move order of a variation he had prepared in depth. He
chose the wrong order and got thrashed by his opponent, who had
remembered the brilliant combination that refuted Anand’s move.

TIME TROUBLE
Competitive games are played with a clock, and thinking time is lim-
ited. The limit is strict and overstepping one’s thinking time means
defeat by “losing on time”, which is not uncommon. An extreme
case is grandmaster Friedrich Sämisch, one of the leading players in
the 1920s, who lost every single game on time in two tournaments
in 1969, when he was 73 years old. The exact time limits vary from
tournament to tournament, but there has been a tendency towards a
decrease of thinking time over the years, partly to reduce the duration
of games and make them more exciting. Currently, many tournaments
use time controls of 90 minutes for the first 40 moves, followed by
 ERRARE HUMANUM EST 83

30 minutes to finish the game, with each player receiving an incre-

ment of 30 seconds after each move.
Before the 1990s, it was common for each player to have 2 hours
30 minutes for 40 moves, and then 1 hour for each sequence of 20
moves. Games used to be interrupted (“adjourned”) after 5 hours
of play, and resumed later in the day or the next day. This break was
meant to allow players to rest and eat. In practice, it also allowed play-
ers and their seconds to analyse the current position. There used to
be an interesting literature describing the excitements and pitfalls of
analysing adjourned games, and about how to do it properly. How-
ever, with the advent of powerful computers, which pretty much
find the optimal sequence of moves for both sides, it became clear
that adjournments were becoming meaningless as the human factor
simply disappeared. The practice of adjourning games progressively
died out in the mid-1990s, with the added advantage that games are
now played in one go, which is much more enthralling from the
spectators’ point of view.
Whatever the exact time controls, it is common for some players
to think for long periods in the opening and at the beginning of the
middle game, when they face critical decisions. As a consequence,
they have little thinking left at the end, and are in time trouble, or in
zeitnot, a German word commonly used by chess players. The exact
definition varies from player to player. For some, having less than 1
minute per move is considered as time trouble; for others, having to
play 10 moves in 5 minutes is nothing special. Players vary consider-
ably as to whether they are likely to be in time trouble. Some players
are in zeitnot in nearly every game, and are used to playing 20 moves
in 2 or 3 minutes. Others systematically avoid any time shortage.
During time trouble, short-term tactics dominate over long-term
strategies, intuition comes before look-ahead search and concrete
ideas take precedence over abstract considerations. Simple heuris-
tics are used, such as “simplify positions by trading pieces off ”,
“restore material balance rather than play for the initiative” and “if
possible, postpone making important decisions until the end of the
zeitnot”. Because decisions have to be made rapidly and calculations
84 E R R A R E H U M A N U M E S T

are superficial, blunders occur often. In general, the advice is to not

be in time trouble. As world champion Alekhine wrote in criticising
one of his moves: “A horrible move, and in my opinion the fact that
White was in time trouble when he made it is no more justification
than the claim of a law breaker that he was drunk when he commit-
ted the crime”.
Alekhine’s comment might be too one-sided. There is actually an
interesting trade-off with respect to decision-making. Not spending
enough time on critical moments in the game might lead to playing
inferior moves, losing the advantage in a superior position or not
defending a difficult position properly. Thus, being in zeitnot might
be a worthwhile price to pay. On the other hand, many players simply
cannot make up their minds, sometimes because they want to find
the optimal move rather than to play a move that is good enough. An
interesting question, which to my knowledge has not been studied,
is whether there is a link between being addicted to time trouble in
chess and procrastination in life outside chess.
Many tragedies have happened in zeitnot, with winning positions
being destroyed in one move. My game against grandmaster Lubomir
Ftacnik in the 1984 Biel tournament is a good example of this. After
a complicated game where my opponent progressively outplayed me,
we were in a mutual time trouble. My opponent made a terrible mis-
take five moves before the time limit, overlooking a simple checkmate
in two moves. His blunder is also a good example of an error caused
by a conditioned reflex. He thought that attacking my rook with a
pawn would force my rook to move away, while in fact moving my
queen created an unstoppable checkmate threat.
This example illustrates the case where being in zeitnot is a ratio-
nal decision. My position had vastly deteriorated in the last 10 moves,
and was clearly lost. Thus, the only practical chance was to create a
situation – a reciprocal time trouble – where the logic of chess could
be perturbed by random factors and by putting psychological pres-
sure on the opponent. While the choice to be in time trouble was
rational from my point of view, it was a mistake from the point of
view of my opponent. I hasten to add that such an analysis is easy
 ERRARE HUMANUM EST 85

with hindsight, but much harder to make during the excitement of

the game. In fact, I found myself in Ftacnik’s unhappy situation in
several of my own games, losing a winning game because of blunders
caused by lack of time.
How to behave when one is in time trouble? Should one try to
exploit or not the opponent’s time trouble? These are important
practical questions, and much ink has been spilled on them. Krogius
answers the first question by advising to keep one’s concentration and
avoiding distracting thoughts. If one has a plan, one should follow
it; else, the best option is to use waiting tactics, since there is simply
not enough time to come up with a sound plan. He also recommends
trying to find the goal of each opponent’s move, even by consciously
asking questions such as “What is the threat?” Finally, he advocates
using auto-suggestion techniques such as verbal commands, the effi-
cacy of which has been documented in sports. As for Kotov, his main
advice is simply to avoid getting into zeitnot. If this not possible, the
main thing is to remain calm and to play as if one were not in time
trouble. Basically, to do the same as normal, just quicker. According
to him, this is what top players are doing.
There is some disagreement about how to handle an opponent’s
zeitnot. On the one hand, both Krogius and Kotov agree on a few
points. One should control one’s nerves, stay calm and evaluate the
position objectively. In particular, one should not get excited or irri-
tated by the opponent’s behaviour. Note that this is not always easy,
as players with few seconds on their clock are sometimes in a state of
extreme agitation if not outright panic. Both also advise against com-
plicating the position for the sake of it, and against speeding up one’s
play to prevent the opponent from thinking during one’s own time, as
this is likely to lead to errors. Where they disagree is whether anything
at all should be done to capitalise on the opponent’s shortage of time.
Krogius recommends calculating a sequence of four or five moves,
which if possible change the nature of the position (e.g. moving from
a middle game to an endgame), and to play them quickly, in the hope
to take one’s opponent unaware. Kotov disagrees, and points out that
this is likely to lead to mistakes on the part of the player who has
86 E R R A R E H U M A N U M E S T

sufficient time. Rather, he endorses world champion Vasily Smyslov’s

advice of leaving the board and coming back only after the opponent
has played a move. Thus, Kotov’s general recommendation is to fully
ignore one’s opponent’s time trouble.

TIREDNESS, OVERCONFIDENCE AND

HABITUATION
This section briefly discusses a few additional sources of errors. An
obvious one is tiredness; after a long and difficult game, or after
several days or weeks of playing in a tournament, one is bound to
become tired. International master Charles Partos, who coached Biel
in the Swiss team championship, used to bring bananas and choco-
late after 4 hours of play to replenish players’ glucose stores and thus
provide energy for the brain – a sound application of nutritional prin-
ciples. The long-term advice to avoid tiredness is to be fit physically,
and many world champions such as Boris Spassky, Bobby Fischer,
Garry Kasparov and Magnus Carlsen practiced several sports and were
in top physical condition when they won the title.
Another standard source of error, not only in chess but in other
sports as well, is overconfidence. With overconfidence, vigilance goes
down, attention becomes relaxed and possible dangers are ignored.
The chess literature contains many examples, even at the top level,
where the proximity of victory led to overconfidence, a decrease in
the quality of play and often errors that turn a win into a draw or,
worse, a loss. Related to overconfidence is the desire to win a game
by a flashy combination rather than by prosaic methods. As objectivity
gets lost, the “brilliant” combination often turns out to be flawed and
the opponent escapes with a lucky draw or even a win.
I have once been victim of an interesting kind of mistake, which I
have not found reported in the chess literature, and which might be
related to the mechanism of habituation. In animal psychology, habitu-
ation is a kind of learning where a response to a stimulus decreases or
even stops after repeated exposure. In my game against former world
champion Boris Spassky at the Reggio Emilia tournament in 1983,
 ERRARE HUMANUM EST 87

I chose a very sharp variation and had the advantage after the opening.
For several moves, I considered a pawn sacrifice in the centre in order
to initiate a direct attack against Spassky’s king. However, the idea
never worked satisfactorily, so I progressively directed my attention
to another central break. Ironically, at the very moment where I gave
up on my initial idea, it would have in fact been the winning move.

HOW TO AVOID ERRORS?

Chess coaches have provided many recommendations for avoiding
errors. Some of them are rather obvious and simply aim to eliminate
the source of the errors I have discussed. For example, one should
be aware of automatisms, control one’s emotions, make sure that
one’s knowledge is up-to-date, relax and get enough sleep before an
important game and avoid time trouble. Some more general recom-
mendations have also been made, which tend to take action against
mind-set effects. A piece of good advice is to look at one’s position
from the point of view of the opponent. This is meant literally: one
should every so often physically stand behind one’s opponent and
have a look at the position. This is actually not bad advice – maybe in
a more figurative way – for many other avenues of life where one is
likely to hold preconceptions.
Kotov encourages players to use Blumenfeld’s rule, named after
Soviet master Benjamin Blumenfeld, who was also a psychologist. The
idea is to make sure that one does not miss the obvious, even after
having spent dozens of minutes analysing a position. After having
selected a move, one should write it down on one’s score sheet, before
actually playing it. (In tournaments, players are required to write down
their moves and their opponents’, as an official record of the game.
This can be useful, for example, to ascertain whether the required
number of moves have been played before the time control.) Kotov
emphasises that care should be taken to write the move neatly and
clearly. He argues that, by doing so, one goes from the future possi-
bilities of the game back to the here and now. A variant of this advice
is to stop looking at the board and stare at some distant object in the
88 E R R A R E H U M A N U M E S T

room. Then, one should look again at the board, with fresh eyes, so
to speak. This should make it possible to look at the position as it is
on the board right now. The advice is now to stare at the position for
about a minute, through the eyes of a novice, and ask very basic ques-
tions such as: is my queen attacked? Is there a direct threat against my
king? According to Kotov, this double-checking procedure drastically
decreases the risk of committing a blunder.
This is very good advice. Having used it in my chess career, I can
vouchsafe that it saved me from an embarrassing oversight more than
once. Unfortunately, one important part of Blumenfeld’s rule is not
possible anymore. The International Chess Federation has changed its
rules, and moves must be written down on the score sheet only after
they have been executed on the board.
 8
PSYCHOLOGICAL WARFARE
AND TRAINING TECHNIQUES

There is a long-standing discussion in the literature as to whether

chess is a sport, a science or an art. Curiously, the obvious fact that it is
a game is rarely mentioned. World champion Emanuel Lasker argued
that it is none of those. Rather, chess is a fight.
Lasker was certainly correct. In fact, this fight can be carried out with
different means: from pure skill (e.g. an attack against a weak pawn), to
subtle psychology (e.g. choosing an opening in which the opponent
feels uncomfortable), to not-so-subtle psychology (e.g. trying to dis-
turb one’s opponent) and finally to outright cheating (e.g. using a com-
puter). Skill is of course the province of the technical chess literature
and is beyond the scope of this book. In the second part of this chapter,
I shall limit myself to discussing ways to improve skill through training.
But the other forms of fight are worth some discussion.

PLAYING THE OPPONENT

As noted by Lasker, chess is a battle between two personalities. Thus,
it is sometimes advantageous to play the opponent rather than objec-
tively strong moves. A tactical trap might be tried against a strategist,
or wild complications might be introduced against a player prefer-
ring clear positions. Playing the opponent is increasingly required to
90 P S Y C H O L O G I C A L WA R F A R E A N D T R A I N I N G

score full points as players’ technical level is getting extremely high,

and simply playing the position might not be enough to win a game
between two players of the same strength.
There is undoubtedly an element of risk in doing so. As Lasker clearly
stated it, playing chess is, to some extent, an application of probability
theory: knowing one’s opponent’s personality, tastes and dislikes makes
it possible to estimate the probability that she would select a given
move. It could be argued that all world champions and most top play-
ers excel in playing the opponent. In Chapter 3, I gave the example of
Kasparov adopting a different style against each of his opponents in his
simultaneous exhibition against the Swiss national team. Chess history
provides some spectacular examples of this style of play.
Let us start with Lasker, the ultimate master of applied chess psy-
chology. In order to win the St. Petersburg tournament in 1914,
Lasker had to beat Capablanca with White. Everybody expected that
Lasker would choose an aggressive or at least a complex opening. To
general surprise, Lasker chose the exchange variation of the Spanish
defence, which had the reputation of being a tame and drawish open-
ing. To boot, he exchanged the queens after five moves, as if he had
no intention of winning. Of course, Lasker knew what he was doing.
The position required Capablanca to play actively to keep the balance,
which he found difficult as he was aiming for a draw. Indeed, Capa-
blanca played passively, and little by little Lasker was able to increase
the mobility of his pieces and build a promising position. In a dif-
ficult situation, Capablanca overlooked an elegant combination, and
had to resign a few moves after.
Another classic example is the overall strategy adopted by Alekh-
ine in his 1927 match against Capablanca. After extensive analysis of
Capablanca’s games, Alekhine came to the conclusion that the Cuban,
whilst a brilliant strategist, was weaker tactically. As consequence, he
decided to systematically double-check Capablanca’s plans for tactical
errors, which did occur in the match. Being prepared, Alekhine was
able to take advantage of them. The Franco-Russian grandmaster made
another brilliant psychological move in this match. With Black, he
used the defensive approach preferred by Capablanca, which was to
 P S Y C H O L O G I C A L WA R F A R E A N D T R A I N I N G 91

simplify positions by trading pieces off and then rely on his technique
to defend them. This approach was new to Alekhine, who in the past
had used more risky and active methods of defence. However, the
psychological surprise turned out to be more important than Alekh-
ine’s relative inexperience with this type of position.

THEORETICAL NOVELTIES
Knowing the “theory of openings” is essential for chess players. It is
not a “theory” in the scientific meaning of the term – a set of mecha-
nisms explaining a body of empirical phenomena. It is not even a set
of principles. Rather, it is the compilation of moves that have been
played in the past and their evaluation: did White or Black have the
advantage, or was the position equal? This body of knowledge has
grown rapidly in the last decades, and it is not uncommon for play-
ers to blitz 20 moves or more in the opening, playing them from
memory. In a paper with Philippe Chassy, we estimated that masters
must have memorised about 100,000 opening moves.
Then comes the much-anticipated moment where the game diverges
from theory: a theoretical novelty. At the amateur level, novelties often
reflect players’ lack of knowledge. By contrast, at the professional level,
many theoretical novelties are the fruit of home preparation, sometimes
weeks or months before the actual game. In the quietness of their living
room, players can spend hours and hours studying a position, moving
the pieces on the board and even asking the advice of their colleagues.
In the last two decades, players have increasingly used computers to
help their investigations, which has multiplied the number of positions
that can be studied by several orders of magnitude. As all this is done
before the start of the game itself, it is legal. The aim of these analyses
is not only to find the move that is objectively the best, but also to find
moves that will confront the opponent with the kinds of positions
they do not like or that are unknown to them. This can occasionally
include inferior positions, especially with Black, with the idea that the
opponent will not be able to find the winning plan in the limited time
allowed by competitive games.
92 P S Y C H O L O G I C A L WA R F A R E A N D T R A I N I N G

The 2008 world championship between Anand and Kramnik

offered several impressive examples of this kind of preparation. For
instance, in the third game, Anand, playing Black, selected a rare but
very sharp defensive system. In the 17th move, he introduced a new
move that sacrificed two pawns for a strong initiative. The line was
highly tactical and risky, but also created the kind of position that
was unpleasant for Kramnik. Indeed, the Russian grandmaster lost
his way in the jungle of complications, and was not able to defend
his position successfully.

INCREASING THE PSYCHOLOGICAL PRESSURE

Moving away from the chessboard, a number of techniques can be
used to get on the opponent’s nerves and disturb him. Whilst most
players behave impeccably, these reprehensible techniques are used
every so often. The options are endless: staring at the opponent inten-
sively (photographs of chess players can be pretty boring, but some
show pretty nasty looks!); arriving late for the game (used sometimes
by Bobby Fischer); distracting the opponent by making noise stirring
up one’s coffee; smoking a strong cigar (one of Lasker’s favourites);
kicking the opponent with one’s feet (occasionally done by Petro-
sian); or wearing a malodorous perfume. Thankfully, the rules of the
International Chess Federation have evolved to avoid some of these
tricks; smoking is now forbidden in the playing hall, and late arrivals
are not always tolerated.
The idea of psychological pressure has sometimes been taken to
surprising extremes. Fine reports that Joseph Henry Blackburne, a
leading player in the second half of the 19th century, threw Steinitz
out of the window after an argument. During the 1984 French cham-
pionship, Gilles Andruet kept distracting Bachar Kouatly with all sorts
of noises, even opening a beer loudly.

Kouatly: “Now, you stop, or I punch your face”.

Andruet: “Go on, I dare you !”
Punch.
 P S Y C H O L O G I C A L WA R F A R E A N D T R A I N I N G 93

OUTRIGHT CHEATING
The last category consists of methods that clearly constitute cheat-
ing. Bobby Fischer accused Soviet grandmasters of such practices:
according to him, they were rapidly drawing amongst themselves
to conserve energy, were consulting during games and were even
fixing games amongst themselves in tournaments. A 2006 study by
Moul and Nye supports Fischer’s view. Using statistical analyses, they
showed that the rate of draws amongst Soviet grandmasters was higher
in important international tournaments than in national tournaments.
Collusion is not uncommon nowadays. The classic example is the
last round of a tournament, where a win would lead to prize money
for one of two players facing each other, but a draw would not. The
temptation is great for the two players to come to an agreement before
the game, with, for example, the player losing the game getting more
money than the winner. On a slightly more bizarre note, a number
of tournaments were simply invented, for example in Rumania and
Ukraine, with no games actually played, so that some players could
win Elo points or be awarded the title of international master or even
grandmaster.
Technology has created new opportunities for cheaters. Several
players have been caught, during a competitive game, using a chess
programme on their smartphone during breaks in the bathroom.
One player, called the “James Bond of chess”, hid a transmitter in
his shoe in order to receive move suggestions sent by a friend. In the
2010 Olympiads, three French players used a sophisticated scheme
for communicating moves. The first player took advantage of live
Internet broadcasting of the tournament to follow the game from his
home and analyse the current position with a computer programme.
He would then text the best move to the second player, who would
communicate it to the player actually playing the game, using his
standing or sitting at various places as a code. The scheme was subtle,
but had a not-so-subtle flaw: they used a mobile phone of the French
Federation, which handed down heavy suspensions to the three play-
ers. To make such technological cheating harder, the International
94 P S Y C H O L O G I C A L WA R F A R E A N D T R A I N I N G

Chess Federation has banned the use of mobile phones and other
electronic devices in playing venues. In the same spirit, tournament
organisers now increasingly add a lag to the transmission of moves
on the Internet.

HOW TO PROGRESS  FROM BEGINNER

TO CANDIDATE MASTER
Once beginners have learnt the basics of the games, which can easily
be done using any one of the many introductory books or material
on the Internet, the question arises: how to progress? This section
deals with training methods suitable up to candidate master level. The
following section deals with the kind of training that more ambitious
players should adopt.
As mentioned in Chapter 4, a number of training methods are
available to dedicated players. First, the basic tactical ideas should
be learnt and mastered. Tactics are essential, as without the ability to
see simple combinations it is impossible to implement any plan suc-
cessfully. In addition, tactics sometimes can help change the course
of a game that was strategically misplayed. Books on tactics typi-
cally consist of descriptions of standard motifs and then numerous
exercises where one has to find the winning sequence of moves. This
is the part of training that is the closest to deliberate practice. Sec-
ond, one should assimilate the foundations of strategy and planning.
A huge number of books and videos are available for teaching this
material. The old classics by Euwe, Tarrasch and Nimzowitsch are still
very useful, as they present the key ideas in a straightforward way. Of
interest are also books focusing on defensive methods or on attacking
methods. Third, endgames should be studied. This is where players
develop their technique: methods for converting a material or strategic
advantage into a win and methods for successfully defending inferior
positions. The great classics by Fine, Averbakh and Dvoretsky are still
good introductions.
Fourth, a fair amount of attention should be devoted to openings.
It is useful to have basic knowledge of most openings, but players
 P S Y C H O L O G I C A L WA R F A R E A N D T R A I N I N G 95

should specialise in particular openings and studying thoroughly.

Here, the advice of a coach is particularly useful, since the amateur
will be bewildered by the huge number of possible openings and
variations. Importantly, studying openings does not solely mean
learning the first moves of the game, but also assimilating typical
strategic ideas and tactical themes. Thus, opening books will spend a
fair amount of time discussing typical middle game positions or even
endgames that often stem from a given opening.
The importance of opening specialisation might appear surpris-
ing, but it is in fact supported by empirical data. In a 2009 study,
Merim Bilalić and colleagues asked players to find the best move in
a given position. The players, who ranged from candidate masters
to grandmasters, specialised either in the Winawer variation of the
French defence or in the Najdorf variation of the Sicilian defence.
The problems came from these two defences and from neutral posi-
tions, which were not typical of any specific opening. With positions
outside the opening in which they specialised, players’ performance
was about 200 Elo points below their performance with familiar posi-
tions. The loss of performance is about one standard deviation in skill,
which is enormous. For example, when facing a position outside
their usual openings, a typical grandmaster (about 2600 Elo points)
would perform at the level of an international master (about 2400
Elo points).
Thus, while part of chess knowledge is general and not related to
any opening, opening-specific patterns and methods are also clearly
important since some level of skill is lost when those cannot be used.
Normally, the advice is to play the openings that one knows well,
especially against stronger players. For a professional player, the night-
mare scenario is to play an obscure tactical opening of which he has
only a hazy memory against a much weaker amateur who obviously
knows all its secrets.
Fifth, one should study the classics: famous historical games and
games played by leading grandmasters, including, of course, world
champions. This will provide a huge source of tactical and strategic
ideas, as well as good examples of how to convert advantages into a
96 P S Y C H O L O G I C A L WA R F A R E A N D T R A I N I N G

win. Sixth, playing competitive games is an essential part of becom-

ing a better player. Finally, it is important to analyse one’s own games
critically, if possible with a coach: What went wrong? What were the
strengths of one’s play? What could be improved?
This literature is essentially applied, in the sense that it is a compila-
tion of principles and methods that have led to successful results. It is
almost never based on scientific research. One of the rare exceptions,
a paper I wrote with Peter Jansen, used template theory (see Chapter 2)
to develop teaching principles. Our approach was guided by some
of the central findings I have discussed in earlier chapters of this book.
These include the limited capacity of attention and short-term mem-
ory, perceptual chunks, selective search and the fact that knowledge is
domain specific. The approach also included more general principles
of learning and memory, such as the importance of processing infor-
mation deeply and ideally from different points of view.
Three educational principles were highlighted in the paper: the
acquisition of knowledge should proceed from simple to complex;
learning is more efficient when the teacher (or a book) clearly identi-
fies the elements to be learnt; and finally, a powerful way of assisting
learning is first to present the material in a simple, even over-simplified
way, and then to return to it several times, each time adding new
information. We called this method an improving spiral. For example,
when studying an opening, a first pass would cover the main ideas of
the opening and commonly played variations. A second pass would
discuss some typical endgames frequent with this opening. The next
pass could address strategic themes suggested by this opening’s pawn
structure. Not only will the student learn important information, but
the multiple passes will also strengthen memory traces and combat
the risk of forgetting. A variant of this is the decomposition method, where
a typical opening position is studied by removing all pieces except
kings and pawns, and the resulting endgame analysed or played in
practice games. Different pieces are progressively added, and again
the resulting endgames studied.
As is clear from my discussion of chess knowledge in Chapter 2,
chess expertise consists in great part of mastering common methods
 P S Y C H O L O G I C A L WA R F A R E A N D T R A I N I N G 97

of play. Thus, while time should be devoted to find-the-best-move

exercises, practice games and analysis, the main ideas should be
explicitly taught and open-ended forms of teaching such as that advo-
cated by the discovery-based learning approach should be avoided.
After all, it took several centuries for very smart people to discover
these ideas!

HOW TO PROGRESS  FROM CANDIDATE

MASTER TO GRANDMASTER
In a sense, training for reaching higher levels is more of the same:
a hearty regime of opening theory, tactics, strategy and endgames.
The literature becomes more advanced and specialised, and at least
an intermediate level of chess understanding is assumed. In addition,
more attention is paid to how to handle games psychologically. At
this stage, increasing use will be made of chess databases and associ-
ated software – highly useful for finding the latest successful ideas
in a given opening – and computers for analysing one’s own games
and opening variations. Players are expected to have assimilated an
opening repertoire, where the variations they are going to play with
White and Black are prepared in advance, studied in great depth and
regularly updated.
The presence of a coach will facilitate several aspects of train-
ing, including the selection of openings, the identification of weak-
nesses and means to correct them and the choice of tournaments.
A coach will also be valuable in preparing a game against specific
opponents, both technically (in particular, the kind of opening varia-
tions to choose) and psychologically (for example, for identifying the
strengths and weaknesses of the opponent).
The literature about how to progress beyond amateur level used to
be dominated by books from the Soviet Union, with some of them
still being classics. Admittedly, these books contained a fair amount
of Marxist propaganda. For example, after Viktor Korchnoi defected
to the West, his games and photographs suddenly disappeared from
chess publications released in the USSR. In addition, the selection of
98 P S Y C H O L O G I C A L WA R F A R E A N D T R A I N I N G

games was clearly biased in favour of Soviet players. Nevertheless,

these books offered excellent overall quality. Classic examples include
Krogius’s book on chess psychology, to which I have referred often in
the previous chapter, Averbakh’s books on endgames and tactics and
Kotov’s book Think like a Grandmaster. This last book provides sophisti-
cated advice about strategy, and rather uniquely, about the proper way
of thinking in chess.
Perhaps the most interesting, and to some extent controversial pro-
posal in Kotov’s book concerns the way one should calculate varia-
tions. Kotov, who was a leading grandmaster in the 1940s and 1950s,
notes in the introduction that for many years he was stuck below
master level, but then in a matter of two years became a grandmaster.
According to him, one of the key reasons for this leap in skill was
his realisation that he was calculating variations very inefficiently. He
would calculate moves in an unsystematic way, jump from one base
move to another, miss important moves and variations, and eventu-
ally end up in time trouble where he would make a blunder. A critical
analysis of his behaviour led him to conclude that one should first
identify candidate moves and then examine them methodically, vis-
iting each branch of the analysis tree only once. Thus, no variation
should be double-checked. However, as we have seen in Chapter 3
when discussing progressive deepening, even top grandmasters do
visit the same lines several times, and their search is far from being
as systematic as computers’ search.
So, what’s going on? Kotov’s method works well in fairly simple
positions, but fails when positions are complex, as even the identifi-
cation of candidate moves is difficult and can indeed be informed by
carrying out search. In addition, progressive deepening is a powerful
means to overcome limits of short-term memory, as it allows some
of the information about the search tree to be encoded in long-
term memory, which thereafter can be used as a virtual short-term
memory (see Chapter 2). However, for training purposes, Kotov’s
method is definitively useful. Also, some of the reasons given for
not doing progressive deepening (e.g. not trusting one’s analysis,
tendency to be in time trouble) are likely to be caused by other
 P S Y C H O L O G I C A L WA R F A R E A N D T R A I N I N G 99

factors rather than the way search is carried out, such as an inability
to make rapid decisions.
Kotov’s book also comes with various exercises for improving
one’s ability to visualise a position and increase depth of search. Most
of these exercises boil down to finding the best move in a given
position, and are certainly useful. Other authors have proposed to
play games blindfold, again to improve one’s visualisation skills. In
the paper with Peter Jansen mentioned earlier, we were rather criti-
cal about this piece of advice. Our argument was that finding good
moves and being able to calculate deep variations was primarily the
consequence of having acquired considerable knowledge, which
allowed players to identify plausible moves rapidly through pattern
recognition, not only in the problem situation but also in the posi-
tions visualised during look-ahead search. Another point was that, if
one has a good understanding of a position, it is often not necessary
to anticipate moves very deeply. However, the paper was written for
players aiming to become masters, and one could argue that, when
one wants to reach higher levels of skill such as grandmaster, playing
blindfold games might help fine-tune one’s calculating skills.
 9
THE MAGIC BULLET?

While most players agree that chess is a fascinating game with endless
possibilities and graced with real aesthetic value, there is disagree-
ment about whether it ultimately benefits or penalises those who play
it. Siegbert Tarrasch famously wrote that “chess, like love, like music,
has the power to make men happy”. Some have highlighted possible
benefits for education. A quick look at the Internet will display many
slogans such as “Chess makes kids smarter” and “Chess improves
mathematics”. Others have directed attention to the potential costs
of playing chess. Orson Wells argued that to destroy a person, you
should teach her chess, and George Bernard Shaw asserted that chess
“is a foolish expedient for making idle people believe they are doing
something very clever when they are only wasting their time”. Rare
are the activities that have simultaneously inspired such extreme views
about their merits and demerits. This chapter deals with the presumed
benefits, and the following chapter with the presumed costs.

INSTRUCTIONAL AND COGNITIVE BENEFITS

One common claim is that playing chess improves children’s perfor-
mance in school, in particular with respect to mathematics and lan-
guage, and generally increases intelligence and other cognitive abilities.
102 T H E M A G I C B U L L E T ?

Numerous websites repeat these claims and some authors use them as
arguments for advertising the practice of chess. In 2001, the US Chess
Trust asked me to review the available evidence, objectively and with-
out any biases. I carried out this review with Guillermo Campitelli.
This is obviously a very important topic for the chess world, and a
positive answer would have significant implications for education. In
fact, it would be a very elegant solution to the difficulties that children
and teenagers currently encounter when studying STEM topics.
This question more generally relates to the issue of transfer: can
abilities acquired in a particular field (e.g. geometry) generalise to
other fields (e.g. science, music) or general abilities (e.g. intelligence,
reasoning)? Broadly speaking, there are two types of transfer. With near
transfer, the two fields overlap, such as geometry and algebra. With far
transfer, there is no or little overlap between the two fields. An example
would be geometry and English. With chess, the strong claim is that
there is far transfer.
Establishing the presence of far transfer and thus the causal role of
chess raises interesting methodological questions. Three groups are
necessary: a treatment group (playing chess, in our case), an active
control group (e.g. playing checkers) and a passive control group (no
treatment). The active control group is needed in order to control
for unspecific factors such as placebo effects and expectations. In
addition, participants should be allocated randomly to the three
groups. This avoids systematic biases, such as the case where all the
more intelligent children are allocated to the chess group whilst the less
intelligent children end up in the other two groups. Measures should
be taken before the experiment to ensure that the groups do not dif-
fer at the outset, and of course after the experiment, to evaluate the
effect of treatment. To measure potential changes, the same variables
should be measured in the pre-test and the post-test. Such a design
in not easy to implement, but is required if one wants to draw strong
conclusions about causality: if the treatment group, but not the other
two groups, improves on the measure(s) of interest, say ability in
mathematics, it is possible to conclude that the effect is specific to
some characteristic of the treatment group.
 THE MAGIC BULLET? 103

Unfortunately, our review turned out to be rather negative.

Although we found many articles extolling the benefits of chess
instruction, very few presented empirical evidence, and of those
that did, even fewer used a proper experimental design. Only three
studies used a design with random assignment and only one study
used an active control group. In addition, the few effects that were
found (specifically: verbal ability and school performance) differed
between the studies and could be due to chance, as a rather large
number of measures were taken. Finally, no study collected data
about the long-term benefits of chess instruction and information
was rarely taken about the characteristics of the teacher. This last
omission is important, as in many experiments teachers were highly
motivated chess players that were convinced of the cognitive benefits
of chess instruction.
A recent review, this time carried out in collaboration with
Giovanni Sala, reached the same conclusions. Rather than a narra-
tive review, we carried out a meta-analysis (see Chapter 4) which
allowed us to quantify some of the outcomes. A positive result was
that the overall quality of the experimental designs had improved
over the years, which was reflected by the fact that many studies
had been published in peer-reviewed journals, a notable step forward
compared to the previous review. However, most of the studies still
lacked an active control group, which as we have seen is a necessary
safeguard against the presence of non-specific effects.
Based on 24 studies, the results indicated a moderate overall effect
size (about one third of a standard deviation). There was a tendency
for the effect size to be stronger with mathematics than with reading
skills. Also, the effects sizes tended to be stronger with longer instruc-
tion. However, caution is required with these results, as no study used
an active control group together with random allocation of partici-
pants to conditions. In fact, Sala conducted two experiments in his
PhD thesis with an active control group (checkers and Go, respec-
tively), and in neither case was there any effect of chess instruction.
In addition, a large study conducted by London’s Institute of Educa-
tion in 100 British schools, and involving just over 4,000 children,
104 T H E M A G I C B U L L E T ?

did not find any effect with respect to performance in mathematics,

literacy and science.
Whilst disappointing, these results are not unique to chess. Sala
and I carried out similar meta-analyses about the presumed effects of
working memory training, music instruction, video-game playing and
exergames (video games coupled with physical activity). The results
were highly consistent: whilst there was evidence for near transfer, far
transfer was very elusive. This pattern of results is well summarised by
the title of Sala’s PhD thesis: Once Upon a Time There Was Far Transfer.
Recently, in part due to the results described in the previous para-
graphs, the aims of the chess-in-school movement seem to have
shifted. Rather than trying to improve general cognitive abilities and
performance in school topics such as mathematics, more modest and
realistic goals are targeted, such as using chess to facilitate integration,
improve social skills and help children make new friends.

OTHER POTENTIAL BENEFITS

Several authors have argued that chess could be used in psychother-
apy. If the patient knows the rules of the game, chess can be used to
initiate a therapeutic relation between the psychotherapist and the
patient, since it offers the advantage of eschewing verbal interac-
tion, which can be problematic with some individuals. In addition,
chess offers a rich symbolism; for example, the promotion of the
pawn into a queen when reaching the other side of the board can be
seen as a child becoming an adult. In particular with psychoanalytical
psychotherapies, this symbolism can be used to express emotions,
aggression and other impulses in an acceptable way. More prosai-
cally, it can be used to improve attention and patience, and teach that
certain rules must be followed. Unfortunately, there is little empirical
evidence on this use of chess. In a short 1949 paper, R. Pakenham-
Walsh reports using chess as occupational therapy in a psychiatric
hospital. Although the article reports the results of chess matches
against local clubs, there is no evaluation of the possible therapeutic
benefits of chess.
 THE MAGIC BULLET? 105

A project in the Netherlands uses chess with children and adults

with an autism spectrum disorder. Chess offers definite advantages
here: clear rules, quiet environment and limited physical and social
contacts. A recent study reports preliminary results suggesting that
playing chess might improve attention deficit hyperactivity disorder
(ADHD) symptoms. After an 11-week chess training programme,
children with ADHD improved on two measures of their symptoms.
A limit of the study is that there was no control group.
It has been proposed that playing chess protects against Alzheimer’s
and other forms of dementia. The rationale is that playing chess keeps
the brain active and builds “cognitive reserve”, which increases the
chances for the brain to function successfully in spite of structural
damage or functional disruption. Unfortunately, preventing dementia
has been notably difficult, and a large number of studies have failed
to provide unequivocal support for the preventative role of a variety
of cognitive activities, including brain training games. A French study
followed 3,675 individuals who were dementia-free at the begin-
ning of the study. It was found that individuals playing games such as
chess, draughts, card games and bingo were 15% less likely to develop
dementia, compared to non-players. However, the effect vanished when
baseline measures of cognition and depression were included in the
analysis. In addition, the study was correlational, with the usual dif-
ficulty in inferring causal links. Thus, since no randomised controlled
trial has been carried out with chess, claims about the protective effects
of chess playing against dementia must be viewed with great caution.
It has also been proposed that chess could help children from
disadvantaged backgrounds, who are at risk of developing drug and
crime problems. A large-scale project in New York City and other US
cities – the motivation behind the Knights of the South Bronx film – found
that playing chess increased reading ability. It is indeed impressive to
see at-risk children enjoying chess and increasing their chances of
completing high school and thus entering universities, against very
serious odds. Similarly, in several countries chess is played in pris-
ons. Some inmates enjoy it and are motivated to improve their skills,
enhance their self-esteem and make new friends.
106 T H E M A G I C B U L L E T ?

These initiatives are inspiring. However, what is lacking for these

and similar claims is clear-cut empirical evidence that they are effec-
tive. In some cases, there are simply no rigorous scientific evaluations.
In addition, the scientific question is to establish whether the benefits
are specific to chess, or whether other interventions would work as
well. For example, with the chess project contributing to the educa-
tion of New York City disadvantaged children, it is possible to imagine
other interventions that would work, as long as they provide alterna-
tives to crime and drugs, increase kids’ self-esteem and offer activities
shared by a group. The answer to these questions would not only be
theoretically interesting, but would also have practical implications.
Before being scientifically confident that the postulated benefits of
chess playing are real, it is necessary to collect reliable data. In many
cases, this can be done with the kind of methodology that I have
discussed with respect to the effects of chess instruction in schools.
Whilst I admire the passion, energy and faith of the persons –
mostly but not exclusively chess players – promoting these initiatives,
I personally do not think that chess is the magic bullet against the
problems facing our society. In my view, chess is a great game provid-
ing much excitement, enjoyment and beauty on its own. There is no
need to justify its practice by alluding to external benefits.
 10
COSTS OF PLAYING CHESS

Observers have not always been sympathetic to chess and some have
even noted a number of possible problems with its practice, which
can be categorised into two groups: non-psychiatric problems and
psychiatric problems. This chapter will address these two categories
in turn, with a focus on professional players.

NONPSYCHIATRIC PROBLEMS
A first obvious cost, which will actually be exacted for any activity
in which one wants to excel, is the huge investment of time that
is required. By necessity, this will leave little time for other activi-
ties. Anecdotal evidence indicates that many chess players aiming to
become professionals have left school early, often immediately after
obligatory school. And among the players who had a longer educa-
tion, many started studying at the university but never graduated.
This lack of completed education puts professional chess players at a
disadvantage if they decide, later in their life, to switch careers.
Although rich in excitement, competitive arousal and fun, a pro-
fessional chess player’s life is also very stressful. There is the obvious
tension caused by competitive games, which last for several hours,
and tournaments, which last for several days. In addition, few players
108 C O S T S O F P L A Y I N G C H E S S

do really well financially. Perhaps the best 20 players in the world

enjoy a high income, and the best 100 players earn a regular income
that provides them with long-term financial security. Weaker profes-
sionals feel the pinch and their often meagre income depends on
winning relatively small amounts of prize money in minor tourna-
ments, teaching, coaching, playing team games and writing books or
columns in newspapers. However, competition is fierce and some of
these activities will interfere with the kind of practice that is necessary
to maintain the level of a strong player.
Chess players often live in rather basic conditions, do a fair amount
of travelling – certainly not in business class – and in general adopt a
mode of life that is not particularly healthy: poor diet, much alcohol
and, for some, smoking. In addition, few players enjoy the security of
health insurance or a pension, which means a lack of control about
their lives.
This stressful style of life is likely to take its toll. Five former world
champions (Morphy, Capablanca, Alekhine, Tal and Petrosian) died
fairly young, at the age of 55 years or younger. A 1969 study by Her-
bert Barry compared strong competitors, weak masters and problem
composers, whose activity entails little competition, with respect to
their longevity. The strong competitors died at 60 years on average,
which was about a decade younger than the individuals in the other
two groups. Being world champion was not a reliable predictor, but
players who had professional interests unrelated to chess lived lon-
ger on average. Given the age of the study, it is a valid question as
to whether the conclusions are still correct, although this seems a
reasonable assumption.
Blindfold chess, especially simultaneous blindfold chess, has
sometimes been singled out as a specific occupational health hazard.
For example, Janos Flesh reported that he needed several weeks of rest
to recover from blindfold exhibitions with dozens of games. Indeed,
such exhibitions were banned in the USSR due to health concerns.
Whilst there is no doubt that simultaneous blindfold chess is a strenu-
ous and tiring mental activity, whether it is dangerous is debatable, in
particular when sufficient attention has been devoted to preparation.
 C O S T S O F P L AY I N G C H E S S 109

PSYCHIATRIC PROBLEMS
A common assumption in newspapers, movies and the chess liter-
ature is that there is a close link between chess and madness. For
example, the 2011 documentary Bobby Fischer against the World played
with the themes of chess, madness and genius. In Chess: A Novel, Ste-
fan Zweig notes that is hard to imagine “a man of intelligence who,
without going mad, again and again, over ten, twenty, thirty, forty
years, applies the whole elastic power of his thinking to the ridiculous
goal of backing a wooden king into the corner of a wooden board!”
The issue of madness is also widespread in the chess psychology
literature, and was certainly a central theme in two books written by
psychoanalyst Reuben Fine, The Psychology of the Chess Player and Bobby
Fischer’s Conquest of the World’s Chess Championship. Understanding the pre-
sumed link between madness and genius is not unique to chess, and
has also been attempted with respect to creativity in music, literature
and the visual arts, in what is known as the “mad genius” hypothesis.
There has been much hype in this literature, but some of the cases
often discussed are rather straightforward, albeit tragic. For exam-
ple, there is much mention of an episode where Steinitz was briefly
admitted in a psychiatric hospital, amongst other things because he
thought he had invented a wireless phone. He was released rapidly
and able to conduct a reasonably normal life after, playing in chess
tournaments. A few weeks before his death, he was again hospital-
ised. One account reports that he wanted to play against God through
electric communication and give Him the advantage of a pawn and
first move. Fine hypothesises that Steinitz was suffering from organic
senile psychosis, with some delusional themes relating to his defeat
in his world championship match against Lasker.
Similarly, Alekhine was known to drink heavily, which led him
to extremes such as being drunk in some of the games of the world
championship against Euwe and urinating in public during a simul-
taneous display. He could also be extremely violent, once destroying
the furniture of his hotel room after losing a game. Whether this
denotes mental illness is debatable, however. This being said, it is
110 C O S T S O F P L A Y I N G C H E S S

likely the case that some top players suffered from psychiatric dis-
eases. The most often discussed in the literature are Paul Morphy and
Bobby Fischer.

Paul Morphy

Paul Charles Morphy was born in 1837. After having learnt to play
chess by watching games between his father and uncle, he was one
of the best players in New Orleans when he was just 9 years old. At
12, he beat Hungarian master Johann Löwenthal 3–0. However, he
did not play much in the following years, devoting himself to his
college and later law studies, which he completed brilliantly. When
he returned to chess at the age of 20, he won all his games in the first
US championship. Since he was not allowed to practice law given his
young age, he travelled to Europe and challenged the best players of
the world, defeating most of them comprehensively, including Adolf
Andersen, the de facto world champion.
After his return to Louisiana in 1859 at the age of 22, he totally
stopped playing chess. He intended to focus on his career as a lawyer,
but the American Civil War (1861–1865) thwarted this plan. He did
not manage to restart his law practice after the end of the war, partly
because his customers were more interested in discussing chess with
him than their legal matters. In spite of this, he lived a comfortable life
thanks to the considerable fortune he had inherited from his family.
He died at the age of 47 from a stroke, after having taken a cold bath
immediately after a long walk on a hot day.
His life was highly organised: daily walk at noon along New
Orleans’ Canal Street, afternoon with his mother and performance
at the opera in the evening. In addition to the fact that he never prac-
ticed as a lawyer after the war, some aspects of his life might denote
symptoms of psychosis. He incorrectly believed that his brother-in-
law had stolen his fortune and wanted to poison him. He methodi-
cally organised his shoes in a half-circle. Occasionally, he would stop
during his daily walk and rather rudely stare at beautiful women. He
also apparently had the habit of walking on his veranda declaiming, in
 C O S T S O F P L AY I N G C H E S S 111

French, “He will plant the flag of Castille on the walls of Madrid with
the cry of the city won and the little king will go away all abashed”.
However, when his family considered placing him in a sanatorium,
Morphy so eloquently defended his legal rights and his sanity that
he was sent home.
According to Fine, these idiosyncrasies support the hypothesis that
Morphy suffered from paranoia. Following Ernest Jones’s earlier anal-
yses (see Chapter 5), he argues that Morphy’s symptoms were caused
by his rivalry with his father, which was sublimated by playing chess.
He notes that Morphy’s international successes started one year after
the early and unexpected death of his father. In addition, Fine takes
Morphy’s refusal to consider chess as a profession as an impossibil-
ity to accept reality and his extreme confidence in his chess abilities
as a sign of exhibitionism. He also proposes that, in his adolescent
years, chess succeeded in protected him from psychosis. When he
stopped playing chess, this defence disappeared, triggering a psy-
chotic regression.
The soundness of this analysis is debatable. Several biographical
details used by Jones and Fine are actually incorrect, as is clear from
reading David Lawson’s very detailed biography of Morphy. Alterna-
tive explanations are not considered, including: the effect of the Civil
War on Morphy’s mental health, the presence of hereditary causes to
his paranoia and the possibility that his decision to stop playing chess
might have been caused by the very bad press that chess had in the
US at the time, being considered as a type of gambling.

Bobby Fischer

Born in 1943, Fischer was predominantly raised by his mother and

rarely saw his father. His childhood was poor and lonely, as his left-
ist mother was politically engaged and paid relatively little interest
to her son. Fischer learnt to play chess at the age of 6, together with
his older sister Joan. He progressed very rapidly. At 14, he became
US champion and at 15 he become the youngest grandmaster in the
world, and indeed the youngest player ever to quality for a candidates’
112 C O S T S O F P L A Y I N G C H E S S

tournament, the penultimate step to win the world title. When he

did not qualify as challenger to the world champion, he (correctly)
accused Russian players of colluding against Western players (see
Chapter 8). Given his chess successes, it is not surprising that he
dropped out of high school at the age of 16, arguing that attending
school was a waste of time.
Fischer’s career was always marred by controversy. For example,
he withdrew from the 1967 Sousse interzonal tournament due to the
poor way the organisation committee dealt with his observance of
the Sabbath, as required by the Seventh-day Adventist Church. When
he left the tournament, he was leading with 8½ points out of 10
games. One year later, he refused to play for the US team in the
Lugano Olympiad, objecting to the unsatisfactory playing conditions.
The real reason might have been that he resented the fact that his fel-
low American grandmaster Samuel Reshevsky was playing first board
for the US.
From 1970 to 1972, he performed at an incredibly high level.
First, he won the Palma de Mallorca interzonal tournament with a
score of 18½ points out of 23 points, 3½ points ahead of the players
finishing joint second. Then, there was 1971, Fischer’s miraculous
year. In quick succession, he destroyed three of the top players in
the world: Taimanov (6–0), Larsen (6–0) and Petrosian (6½ – 2½).
The chess world had never seen anything like this. This extraordinary
string of victories gave Fischer the right to challenge world champion
Boris Spassky.
The match was preceded by complex negotiations, including
about its location, which ultimately was agreed to be Reykjavik. Even
the day before the official starting date of the match, nobody knew
whether Fischer would play. Finally, apparently after being convinced
by his friend Anthony Saidy, a chess international master and a medi-
cal doctor, he agreed to compete. After losing the first game due to
a strange oversight in a basic endgame, he lost the second game by
forfeit – as a protest against the noise made by the cameras in the
playing room. After this shaky beginning, Fischer started to outclass
Spassky with his brilliant play, plethora of new ideas in the openings
 C O S T S O F P L AY I N G C H E S S 113

and continuous psychological pressure. Fischer eventually won the

match by 12½ to 8½. After his victory, he withdrew from chess and
lived a reclusive life. In particular, he declined to defend his title in
1975, as no agreement about the match conditions could be reached.
It was only in 1992 that he played again, in a controversial match
against Spassky that was held in Sveti Stefan and Belgrade. Yugosla-
via was at the time under a United Nations embargo, which the US
endorsed, and Fischer was thus breaking American law by playing
there. It is likely that he did it due to financial pressure: the purse of
the match was $5 million, with $3.35 million going to winner. He
won the match 17½–12½. It is generally accepted that Fischer, while
still playing at a high level, was clearly weaker than the then world
champion Garry Kasparov.
Fischer’s violation of American law meant that he could not return
to the US. He lived in exile in various countries, including Hungary,
the Philippines and Japan, until his death in 2008 in Reykjavik, the
city where he became world champion. He died from renal failure,
after refusing medical treatment from fear of being poisoned, at the
symbolic age of 64 (the number of squares on a chessboard).
In his book about the Fischer-Spassky match, Reuben Fine dis-
cussed Fischer’s psychology at great length. For him, Fischer was
“a troubled human being”, suffering from “obvious personal prob-
lems”. Fine and others argued that Fischer displayed several symptoms
indicative of paranoia. As evidence for this diagnosis, the following
traits are mentioned: his hatred of Jews; his denial of the Holocaust;
his accusations that Soviet grandmasters cheated against him; his con-
stant complaints about playing conditions, including lighting, noise
and TV cameras; and with one exception, his absence of romantic
interaction with women. The last two points merit some comment.
Whilst Fischer did complain about playing conditions, he was often
justified and was in fact credited with vastly improving the prize
money offered to professional chess players. With respect to Fischer’s
relationships with women, it should be noted that he did have several
of them after winning the world championship, including getting
married to Miyoko Watai, a Japanese international master.
114 C O S T S O F P L A Y I N G C H E S S

Fine makes two interesting points about Fischer’s mental prob-

lems. The first is that, despite his troubled personality and childish
behaviour, he was sometimes able to react with maturity and a strong
sense of reality to difficult situations. This was the case after the losses
incurred in the first two games in his match against Spassky, after
which he played superb chess. The second is that, to some extent,
chess might have served as the best therapy for Fischer. When he
stopped playing after the world championship, his paranoid symp-
toms increased.

Is there anything to explain?

These hypotheses about the link between chess and madness are
intriguing, but also controversial. A first problem is that the quality
of the data is poor – anecdotes, personal observations by colleagues,
press reports, etc. It is noticeable that Fine, although he personally
knew several of the champions he discussed in his books, never car-
ried out a formal and detailed clinical evaluation of any of them.
Second, there is no consideration of base rates. Noting the pres-
ence of paranoid personalities, for example, is one thing, but before
speculating on any link between genius and madness in chess, one
needs to establish that the prevalence of paranoid personalities (or
other psychiatric illnesses) in chess is higher than in the general
population. Given that the data span two centuries, this is obviously
difficult to do, as the prevalence is likely to have changed during this
period of time. Similarly, geographical and cultural differences should
be taken into account.
Third, and related to the previous point, it is obvious that some of
the world champions have lived in extremely difficult conditions –
think of the financial duress faced by Steinitz most of his life, and
Morphy’s experience of the Civil War. Psychoanalytical analyses do
not take this into account.
Finally, many of the analyses presented in this chapter are based
on psychoanalysis, which has been discredited scientifically and only
plays a minor role today in psychology and psychiatry. Specifically,
 C O S T S O F P L AY I N G C H E S S 115

psychoanalytical theories tend to be vague and ill-defined, and it is

thus hard if not impossible to test them. Currently, life experiences
in childhood and adulthood as well as genetics are considered as
more likely causes of mental problems than the kind of explanations
adduced by Jones and Fine. In the case of Fischer, a genetic explana-
tion seems very plausible, given that both his parents showed para-
noid symptoms.
In general, when considering the current leading grandmasters –
say, the top 100 players – there is little, if any, sign of psychiatric
illness. To be sure, different personality types are represented, but
obviously this does not imply pathological personalities.
 11
ENDGAME

We have played a long game and have now reached its final stage. It is
time to draw together the different strands of this book.
Researchers now have a fairly good understanding of the topics
covered in the first four chapters. We know that perception is cen-
tral to expertise in chess, knowledge is mostly stored in long-term
memory as perceptual chunks linked to possible actions and search is
highly selective. We also know that skill in chess mostly finds it origin
in a combination of innate factors and dedicated practice. The topics
in the second half of the book have been less researched, and less is
known about them. They also tend to be more applied and harder to
study experimentally.
One theme that has been present in nearly all chapters is that of
bounded rationality. Developed in the 1950s by Herbert Simon –
whom we have met in several chapters of this book – bounded ratio-
nality proposes that humans make decisions that are good enough,
but not optimal, because of the limits imposed by their cognitive
resources. On the one hand, chess grandmasters choose remarkably
good moves: they are nearly always within the first three best moves
selected by the top computer engines, and often match the best one.
On the other hand, this book has presented substantial empirical
evidence that their rationality is limited: their search is extremely
118 E N D G A M E

selective, covering only a minute portion of the possible search space;

they sometimes commit small errors and more rarely are victims of
terrible blunders; they cannot transfer their knowledge from chess to
other domains; and they can succumb to psychological ruses taking
advantage of shortcomings of their style. More controversially, other
aspects of chess psychology can be taken as evidence for bounded
rationality: the need to cheat for winning games and the suboptimal
contact with reality displayed by some players.
In the end, chess has been true to its reputation of drosophila of
cognition. It has much contributed to our understanding of human
cognition, not the least because it is both a closed and complex world.
This will continue to be the case in the future, with new methods and
new scientific questions.
 FURTHER READING

Barry, H. (1969). Longevity of outstanding chess players. Journal of Genetic Psychol-

ogy, 115, 143–148.
This article provides data showing that leading chess players have a
shorter life expectancy than weaker players.
Bilalić, M., McLeod, P., & Gobet, F. (2010). The mechanism of the Einstellung
(set) effect: A pervasive source of cognitive bias. Current Directions in Psycho-
logical Science, 19, 111–115.
This article provides an overview of the experiments carried out on set
effects in chess. Particularly revealing is the experiment using eye-move-
ment recording.
Bilalić, M., Smallbone, K., Mcleod, P., & Gobet, F. (2009). Why are (the best)
women so good at chess? Participation rates and gender differences in
intellectual domains. Proceedings of the Royal Society B, 276, 1161–1165.
Based on a statistical analysis, this article argues that participation rates
should be taken into account before other explanations for the presence
of gender differences in chess are considered.
Blasco-Fontecilla, H., et al. (2016). Efficacy of chess training for the treatment
of ADHD: A prospective, open label study. Revista de Psiquiatría y Salud Mental,
9, 13–21.
This study uses chess for reducing symptoms of ADHD.
120 F U R T H E R R E A D I N G

Burgoyne, A. P., Sala, G., Gobet, F., Macnamara, B. N., Campitelli, G., & Hambrick,
D. Z. (2016). The relationship between cognitive ability and chess skill: A
comprehensive meta-analysis. Intelligence, 59, 72–83.
This meta-analysis clearly shows that expertise in chess is related to
intelligence.
Campitelli, G., & Gobet, F. (2004). Adaptive expert decision making: Skilled
chess players search more and deeper. Journal of the International Computer
Games Association, 27, 209–216.
In the experiment reported in this paper, participants had to solve fiend-
ishly complicated problems. The verbal protocols revealed very clear-cut
skill differences in the amount of search that was carried out.
Campitelli, G., Gobet, F., & Bilalić, M. (2014). Cognitive processes and devel-
opment of chess genius: An integrative approach. In D. K. Simonton
(Ed.), The Wiley handbook of genius (pp. 350–374). Chichester, UK: Wiley-
Blackwell.
This book chapter describes a model of the interaction of practice and
talent in chess and derives predictions through mathematical simulations.
Charness, N. (1976). Memory for chess positions: Resistance to interference.
Journal of Experimental Psychology: Human Learning and Memory, 2, 641–653.
This article presents several experiments showing that memory for briefly
presented chess positions is unaffected by interpolated tasks. This result
challenges chunking theory.
Charness, N., Reingold, E. M., Pomplun, M., & Stampe, D. M. (2001). The per-
ceptual aspect of skilled performance in chess: Evidence from eye move-
ments. Memory & Cognition, 29, 1146–1152.
This article provides interesting data on the role of eye movements when
chess players try to find the best move in a position.
Chassy, P. & Gobet, F. (2010). Speed of expertise acquisition depends upon
inherited factors. Talent Development and Excellence, 2, 17–27.
This article proposes a genetic hypothesis that accounts for inter-individ-
ual differences in the acquisition of expertise.
Dartigues, J. F., et al. (2013). Playing board games, cognitive decline and demen-
tia: A French population-based cohort study. BMJ Open, 3.
This is a large study looking at the possible benefits of playing board
games and other games for reducing the risk of dementia.
De Groot, A. D. (1978). Thought and choice in chess (2nd English ed.; 1st Dutch ed. in
1946). The Hague: Mouton Publishers.
 F U RT H E R R E A D I N G 121

De Groot’s magnum opus is a must for anybody interested in chess psy-

chology. The book focuses on chess players’ thinking processes when
they choose a move, but also discusses the role of perception and
memory.
De Groot, A. D., Gobet, F., & Jongman, R. W. (1996). Perception and memory in chess:
Heuristics of the professional eye. Assen: Van Gorcum.
This systematic study of the role of perception in chess provides detailed
analyses of eye-movement data. It also discusses findings and theories on
memory. In the last chapter, the first two authors defend their different
views about the nature of intuition in chess.
Dreyfus, H. L., & Dreyfus, S. E. (1986). Mind over machine. New York, NY: Free Press.
This book presents the influential five-stage theory of intuition. Chess is
used repeatedly to support the authors’ argument.
Ericsson, K. A., Krampe, R. T., & Tesch-Römer, C. (1993). The role of deliberate
practice in the acquisition of expert performance. Psychological Review, 100,
363–406.
This is the key paper on deliberate practice. It focuses on music, but also
uses chess as a source of evidence.
Euwe, M. (1978). The development of chess style. New York, NY: David McKay.
This book presents the thesis that the development of players’ individual
style parallels the historical development of styles.
Ferrari, V., Didierjean, A., & Marmèche, E. (2006). Dynamic perception in chess.
Quarterly Journal of Experimental Psychology, 59, 397–410.
This article presents two fascinating experiments showing that anticipa-
tory schemas might lead to errors.
Fine, R. (1967). The psychology of the chess player. New York, NY: Dover.
This is the classic but controversial application of Freudian psychoanalysis
for explaining chess players’ unconscious motivations.
Fine, R. (1973). Bobby Fischer’s conquest of the world’s chess championship:The psychology and
tactics of the title match. New York, NY: David McKay.
This book discusses the application of psychoanalytical theory to the
match of the century.
Gobet, F. (2016). Understanding expertise: A multidisciplinary approach. London: Palgrave.
This is a comprehensive discussion of research into expertise. Since chess
has much contributed to the topic, it is covered in some detail, not only
from the point of view of psychology, but also from the point of views of
neuroscience, philosophy and artificial intelligence.
122 F U R T H E R R E A D I N G

Gobet, F., & Campitelli, G. (2006). Education and chess: A critical review. In
T. Redman (Ed.), Chess and education: Selected essays from the Koltanowski conference (pp.
124–143). Dallas, TX: Chess Program at the University of Texas at Dallas.
This book chapter reviews the research using chess for improving cogni-
tive abilities and educational achievement. It provides a detailed discussion
on the ideal methodology to use for carrying out this type of research.
Gobet, F., & Campitelli, G. (2007). The role of domain-specific practice, handed-
ness and starting age in chess. Developmental Psychology, 43, 159–172.
This is one of the first papers to challenge the deliberate practice frame-
work with the kind of empirical data used by this approach. Its contribu-
tion was also to combine data on deliberate practice with data on talent.
Gobet, F., & Chassy, P. (2009). Expertise and intuition: A tale of three theories.
Minds & Machines, 19, 151–180.
This article presents a critique of Simon and Chase’s and Dreyfus and
Dreyfus’s theories of intuition, and proposes a new theory, based on tem-
plate theory (see Chapter 2).
Gobet, F., de Voogt, A. J., & Retschitzki, J. (2004). Moves in mind. Hove, UK: Psy-
chology Press.
This book presents a systematic review of the psychology of board games,
with particular attention to chess.
Gobet, F., & Jansen, P. J. (2006). Training in chess: A scientific approach. In
T. Redman (Ed.), Chess and education: Selected essays from the Koltanowski conference
(pp. 81–97). Dallas, TX: Chess Program at the University of Texas at Dallas.
This book chapter uses results in cognitive psychology to develop prin-
ciples of chess instruction.
Gobet, F., Lane, P. C. R., Croker, S., Cheng, P. C. H., Jones, G., Oliver, I., & Pine,
J. M. (2001). Chunking mechanisms in human learning. Trends in Cognitive
Sciences, 5, 236–243.
This is a review showing the general role of chunking mechanisms in
human cognition, beyond chess. It provides a brief presentation of the
CHREST model.
Gobet, F., & Simon, H. A. (1996). The roles of recognition processes and look-
ahead search in time-constrained expert problem solving: Evidence from
grandmaster level chess. Psychological Science, 7, 52–55.
This paper analyses world champion Garry Kasparov’s simultaneous
exhibitions against national teams and the team of Hamburg. It takes
 F U RT H E R R E A D I N G 123

Kasparov’s remarkable results as support for the role of pattern recogni-

tion in chess skill and expertise in general.
Gobet, F., & Simon, H. A. (1996). Templates in chess memory: A mechanism for
recalling several boards. Cognitive Psychology, 31, 1–40.
This article describes experiments challenging chunking theory and pres-
ents a revision of this theory.
Guid, M., & Bratko, I. (2006). Computer analysis of chess champions. ICGA Jour-
nal, 29, 65–73.
This fascinating but also contentious article uses computer analyses to
study the style of world champions.
Halpern, D. F. (2013). Sex differences in cognitive abilities (4th ed.). New York, NY:
Psychology Press.
Diane Halpern is the world expert on gender differences in cognition. The
book systematically reviews the empirical evidence and the theories that
have been proposed to explain it. Recommended if you are interested in
gender differences.
Hearst, E., & Knott, J. (2009). Blindfold chess: History, psychology, techniques, champions,
world records and important games. Jefferson, NC: McFarland.
This book presents a systematic review of the literature on blindfold
chess.
Holding, D. H. (1985). The psychology of chess skill. Hillsdale, NJ: Erlbaum.
In this book, Holding criticises Simon and Chase’s chunking theory and
their emphasis on pattern recognition, and argues rather that the ability
to look-ahead is the main factor in chess skill.
Kotov, A. (1971). Think like a grandmaster. London: Batsford.
This classic in the chess literature provides advice for ambitious players
about how to become a grandmaster. The emphasis is on developing dis-
cipline for calculating variations.
Krogius, N. (1976). Psychology in chess. London: R. H. M Press.
This book is another classic in the chess literature. Nicolai Krogius
describes a large number of competitive games where psychology
played an important role. The book rarely refers to the scientific litera-
ture, but the author demonstrates a very good understanding of human
psychology.
Lawson, D. (1976). The pride and sorrow of chess. New York, NY: David McKay.
This is the definitive biography of Paul Morphy.
124 F U R T H E R R E A D I N G

Maass, A., D’Ettole, C., & Cadinu, M. (2008). Checkmate? The role of gender
stereotypes in the ultimate intellectual sport. European Journal of Social Psychol-
ogy, 38, 231–245.
The article argues that male domination in chess is due to gender
stereotypes.
Moul, C. C., & Nye, J. V. (2006). Did the Soviets collude? A statistical analy-
sis of championship chess 1940–64. Retrieved from http://dx.doi.
org/10.2139/ssrn.905612.
An interesting statistical analysis that supports Bobby Fischer’s claim that
Soviet grandmasters essentially cheated against Western players.
Murray, H. J. R. (1952). A history of chess (reprint of the 1913 ed.). New York, NY:
Skyhorse Publishing Press.
First published in 1913, this is the definitive reference on the history of
chess.
Polgár, L. (2017). Raise a genius! (Original ed. in Hungarian, 1989). Translation
and copyright: Gordon Tisher.
In this book, László Pogar presents his views on education and describes
the “experiment” he and his wife carried out on their three daughters.
For many years, the book was available only in Hungarian and Esperanto.
It can be downloaded at http://slatestarcodex.com/Stuff/genius.pdf.
Saariluoma, P. (1992). Error in chess: The apperception-restructuring view. Psy-
chological Research, 54, 17–26.
This paper describes elegant studies where errors are induced
experimentally.
Saariluoma, P. (1995). Chess players’ thinking: A cognitive psychological approach. London:
Routledge.
This book discusses the psychology of chess, with a focus on the author’s
research. Experiments on blindfold chess are discussed in some detail.
Sala, G., & Gobet, F. (2016). Do the benefits of chess instruction transfer to
academic and cognitive skills? A meta-analysis. Educational Research Review,
46–57.
This meta-analysis statistically combines the results of 24 studies.
Silver, D., et al. (2016). Mastering the game of Go with deep neural networks
and tree search. Nature, 529, 484–489.
AlphaGo, which is described in this paper, revolutionised the way com-
puters play Go.
 F U RT H E R R E A D I N G 125

Simon, H. A., & Chase, W. G. (1973). Skill in chess. American Scientist, 61, 393–403.
In this classic paper, Simon and Chase provide an overview of the seminal
research they carried out on chess.
Stafford, T. (2018). Female chess players outperform expectations when playing
men. Psychological Science, 29, 429–436.
Using data from a large database, this article argues that women do not
suffer from any stereotype threat, but rather play better than expected
when facing men.